Merge main

This commit is contained in:
Chris Tate
2026-10-22 23:23:00 -05:00
217 changed files with 7438 additions and 537 deletions
+12 -7
View File
@@ -773,17 +773,22 @@ jobs:
run: >-
pnpm exec vitest run tests/harness/worktree-files.test.ts
--testNamePattern "workspace reset keeps Linux Sandbox paths POSIX"
# The Node host checks with tsc-ts, the default, over the named pipe it
# serves on Windows, and with tsgo.
- name: Windows scriptc run smoke
shell: pwsh
run: |
zig version
$output = & node packages/cli/dist/main.js run tests/corpus/001-hello.ts --backend llvm
if ($LASTEXITCODE -ne 0) {
exit $LASTEXITCODE
}
$actual = ($output -join "`n").Trim()
if ($actual -ne "hello world") {
throw "unexpected scriptc output: $actual"
foreach ($checker in @("tsc-ts", "tsgo")) {
$env:SCRIPTC_CHECKER = $checker
$output = & node packages/cli/dist/main.js run tests/corpus/001-hello.ts --backend llvm
if ($LASTEXITCODE -ne 0) {
exit $LASTEXITCODE
}
$actual = ($output -join "`n").Trim()
if ($actual -ne "hello world") {
throw "unexpected scriptc output with ${checker}: $actual"
}
}
- name: Windows PE subsystem selection
env:
+1 -1
View File
@@ -206,7 +206,7 @@ With `--sparse`, the profile leaves out the functions the runs never reached. Th
<dd>A runtime pack directory to use instead of an installed <code>@scriptc/runtime-&lt;target&gt;</code> package.</dd>
<dt><code>SCRIPTC_CHECKER</code></dt>
<dd>The TypeScript 7 API server that checks the program: <code>tsc-ts</code> (the default), which scriptc includes, or <code>tsgo</code>, from the <code>typescript</code> package scriptc also includes. Both give the same diagnostics and generated code. On Windows the default is <code>tsgo</code>, because tsc-ts's API server does not yet accept the named pipe scriptc connects over there.</dd>
<dd>The TypeScript 7 API server that checks the program: <code>tsc-ts</code> (the default), which scriptc includes, or <code>tsgo</code>, from the <code>typescript</code> package scriptc also includes. Both give the same diagnostics and generated code.</dd>
</dl>
```console
+1 -1
View File
@@ -40,7 +40,7 @@
},
"dependencies": {
"@scriptc/runtime": "workspace:*",
"tsc-ts": "0.0.7",
"tsc-ts": "0.0.8",
"typescript": "7.0.2",
"typescript5": "npm:typescript@5.9.3"
},
@@ -57,10 +57,11 @@ test("only empty literals stored where the module appends construct small", () =
expect(found.has(idle)).toBe(false);
expect(found.has(returned)).toBe(false);
// Only `fill` builds arrays.
// Only `fill` builds arrays. They may hold int32 slots (element kind 7),
// which does not change the choice of constructor.
const llvm = emitLlvmModule(module(body, ["xs", "ys"]));
expect(llvm.match(/call ptr @scr_arr_new_small\(i32 0\)/g)).toHaveLength(2);
expect(llvm.match(/call ptr @scr_arr_new\(i32 0, i64 0\)/g)).toHaveLength(2);
expect(llvm.match(/call ptr @scr_arr_new_small\(i32 [07]\)/g)).toHaveLength(2);
expect(llvm.match(/call ptr @scr_arr_new\(i32 [07], i64 0\)/g)).toHaveLength(2);
});
test("a binding with the same id in another function does not count", () => {
@@ -33,7 +33,7 @@ import {
import type { LlValue, LlvmEmitterContext } from "./expr-context.js";
import { compactRefLoadLines, compactRefStoreLines } from "./classes.js";
import { elemAccess } from "./shapes.js";
import { exactInteger, widenInteger } from "./integer-values.js";
import { indexInteger } from "./integer-values.js";
import { integerOrNaNIndex } from "./wrapping-sums.js";
const SCR_ARR_VALUE = 1;
@@ -99,14 +99,17 @@ function emitDenseGuard(
): DenseGuard {
const B = host.B;
const size = host.sizeType;
const integer = exactInteger(host, index, indexExpr);
const nanIndex = integer ? null : integerOrNaNIndex(host, index, indexExpr);
const integer = indexInteger(host, index, indexExpr);
const nanIndex = integer === null ? integerOrNaNIndex(host, index, indexExpr) : null;
let wide: string;
let integral: string | null = null;
if (integer) {
// A proven integer needs no double round-trip. Negative signed values
// compare above any length or capacity as unsigned 64-bit integers.
wide = widenInteger(host, integer);
if (integer !== null) {
// A proven integer, or a wide view that equals the double at every
// index below a length, needs no double round-trip. Negative signed
// values compare above any length or capacity as unsigned 64-bit
// integers, and every index that fails the comparisons takes the
// double's path.
wide = integer;
} else if (nanIndex) {
// An exact integer or NaN is an index exactly when it is ordered.
wide = nanIndex.wide;
@@ -239,9 +242,10 @@ function statePointer(
return statePtr;
}
/** How an array's dense slots hold its elements: the access kind, or
* `cref` for references to compact class instances (SCR_ELEM_CREF: a 4-byte
* slot holding the address over 8; LlEmitter.compactArrayElem). */
/** How an array's dense slots hold its elements: the access kind (`i32`
* for int32 arrays, ir/int32-arrays.ts), or `cref` for references to compact
* class instances (SCR_ELEM_CREF: a 4-byte slot holding the address over 8;
* LlEmitter.compactArrayElem). */
type SlotKind = ArrayAccess | "cref";
function slotKind(host: LlvmEmitterContext, acc: ArrayAccess, element: IrType): SlotKind {
@@ -249,14 +253,14 @@ function slotKind(host: LlvmEmitterContext, acc: ArrayAccess, element: IrType):
}
/** The dense slot type: boolean arrays store one byte per element, compact
* references four, every other element kind an 8-byte slot. */
* references and int32 arrays four, every other element kind an 8-byte slot. */
function slotType(kind: SlotKind): "i8" | "i32" | "i64" {
return kind === "bool" ? "i8" : kind === "cref" ? "i32" : "i64";
return kind === "bool" ? "i8" : kind === "cref" || kind === "i32" ? "i32" : "i64";
}
/** Bytes per dense slot (scr_arr_slot_width). */
function slotWidth(kind: SlotKind): number {
return kind === "bool" ? 1 : kind === "cref" ? 4 : 8;
export function slotWidth(kind: SlotKind): number {
return kind === "bool" ? 1 : kind === "cref" || kind === "i32" ? 4 : 8;
}
/** Bytes per dense slot of an array of `element` (scr_arr_slot_width). */
@@ -264,6 +268,17 @@ export function denseSlotWidth(host: LlvmEmitterContext, element: IrType): numbe
return slotWidth(slotKind(host, elemAccess(element), element));
}
/** The runtime accessor family of an access: int32 arrays answer through
* the double-valued entry points, which convert their slots. */
export function runtimeAccess(acc: ArrayAccess): "f64" | "bool" | "ref" {
return acc === "i32" ? "f64" : acc;
}
/** The LLVM type of an access's values at the emitter's interface. */
function valueType(acc: ArrayAccess): "double" | "i1" | "ptr" {
return acc === "f64" || acc === "i32" ? "double" : acc === "bool" ? "i1" : "ptr";
}
function valuePointer(
host: LlvmEmitterContext,
guard: Pick<DenseGuard, "offset" | "data">,
@@ -285,7 +300,13 @@ function loadState(host: LlvmEmitterContext, guard: DenseGuard): string {
return state;
}
export type ArrayAccess = "f64" | "bool" | "ref";
/** The ScrElemKind of int32 arrays (scr_array.h). */
export const SCR_ELEM_I32 = 7;
/** How a dense array's slots are read and written: doubles, booleans,
* references, or the int32 slots of SCR_ELEM_I32 arrays (ir/int32-arrays.ts),
* whose values are doubles at the emitter's interface. */
export type ArrayAccess = "f64" | "bool" | "ref" | "i32";
/** Before a self-append (`a[i] = a[i] + suffix`) replaces a reference
* element: when `index` lies in the hole-free prefix and the slot still
@@ -341,15 +362,33 @@ export function emitDenseArrayGet(
retain: boolean,
proven = false,
): string {
return emitDenseArrayRead(host, array, index, indexExpr, acc, element, getter, retain, proven)
.value;
}
/** emitDenseArrayGet, with the int32 of an int32 array's element: the
* slot itself on the dense path, the runtime getter's double converted
* (exactly: the element is an int32) on the other. */
export function emitDenseArrayRead(
host: LlvmEmitterContext,
array: string,
index: LlValue,
indexExpr: IrExpr | undefined,
acc: ArrayAccess,
element: IrType,
getter: string,
retain: boolean,
proven = false,
): { value: string; int32: string | null } {
const B = host.B;
const ty = acc === "f64" ? "double" : acc === "bool" ? "i1" : "ptr";
const ty = valueType(acc);
const ret = acc === "bool" ? "zeroext i1" : ty;
const kind = slotKind(host, acc, element);
host.declare(`declare ${ret} @${getter}(ptr, double)`);
if (acc === "ref" && retain && !isRefCounted(element)) {
const t = B.tmp();
B.line(`${t} = call ${ty} @${getter}(ptr ${array}, double ${index.name})`);
return t;
return { value: t, int32: null };
}
const slow = B.newLabel("arr.get.slow"),
join = B.newLabel("arr.get.join"),
@@ -380,10 +419,16 @@ export function emitDenseArrayGet(
B.startBlock(valueLabel);
const ptr = valuePointer(host, guard, kind);
let fast: string;
let fastInt: string | null = null;
let fastLabel = valueLabel;
if (acc === "f64") {
fast = B.tmp();
B.line(`${fast} = load double, ptr ${ptr}${host.fieldAliasAttachment(ptr)}`);
} else if (acc === "i32") {
fastInt = B.tmp();
fast = B.tmp();
B.line(`${fastInt} = load i32, ptr ${ptr}${host.fieldAliasAttachment(ptr)}`);
B.line(`${fast} = sitofp i32 ${fastInt} to double`);
} else if (acc === "bool") {
const raw = B.tmp();
fast = B.tmp();
@@ -410,11 +455,21 @@ export function emitDenseArrayGet(
const slowValue = B.tmp();
const attrs = acc === "ref" && retain ? " cold" : READ_FALLBACK;
B.line(`${slowValue} = call ${ty} @${getter}(ptr ${array}, double ${index.name})${attrs}`);
let slowInt: string | null = null;
if (fastInt !== null) {
slowInt = B.tmp();
B.line(`${slowInt} = fptosi double ${slowValue} to i32`);
}
B.br(join);
B.startBlock(join);
const result = B.tmp();
B.line(`${result} = phi ${ty} [ ${fast}, %${fastLabel} ], [ ${slowValue}, %${slow} ]`);
return result;
let int32: string | null = null;
if (fastInt !== null && slowInt !== null) {
int32 = B.tmp();
B.line(`${int32} = phi i32 [ ${fastInt}, %${fastLabel} ], [ ${slowInt}, %${slow} ]`);
}
return { value: result, int32 };
}
/** `scr_arr_get_number`: a numeric element, or NaN for every missing,
@@ -425,13 +480,14 @@ export function emitDenseArrayNumber(
index: LlValue,
indexExpr: IrExpr | undefined,
proven = false,
acc: "f64" | "i32" = "f64",
): string {
return emitDenseArrayGet(
host,
array,
index,
indexExpr,
"f64",
acc,
{ kind: "f64" },
"scr_arr_get_number",
false,
@@ -486,7 +542,9 @@ export function emitDenseArrayState(
* value's reference. A dense index stores the slot, marks it present, and
* extends the length when it lands at or past it; a replaced present
* reference is released after the new edge is published (a release can
* collect cycles). Every other index takes the unchanged runtime call. */
* collect cycles). Every other index takes the unchanged runtime call.
* `int32`, for an int32 array, is the value as an i32 when the caller has
* it at hand. */
export function emitDenseArraySet(
host: LlvmEmitterContext,
array: string,
@@ -496,11 +554,12 @@ export function emitDenseArraySet(
element: IrType,
value: string,
proven = false,
int32: string | null = null,
): void {
const B = host.B;
const ty = acc === "f64" ? "double" : acc === "bool" ? "i1" : "ptr";
const ty = valueType(acc);
const kind = slotKind(host, acc, element);
const setter = `scr_arr_set_${acc}`;
const setter = `scr_arr_set_${runtimeAccess(acc)}`;
host.declare(`declare void @${setter}(ptr, double, ${acc === "bool" ? "i1 zeroext" : ty})`);
const call = (attrs: string): void =>
B.line(`call void @${setter}(ptr ${array}, double ${index.name}, ${ty} ${value})${attrs}`);
@@ -508,6 +567,13 @@ export function emitDenseArraySet(
call("");
return;
}
// An int32 array's stores are proven int32s (ir/int32-arrays.ts); both
// dense paths below store the converted value.
let int32Value: string | null = int32;
if (acc === "i32" && int32Value === null) {
int32Value = B.tmp();
B.line(`${int32Value} = fptosi double ${value} to i32`);
}
const slow = B.newLabel("arr.set.slow"),
join = B.newLabel("arr.set.join"),
filledLabel = B.newLabel("arr.set.filled");
@@ -526,6 +592,8 @@ export function emitDenseArraySet(
const storeValue = (valuePtr: string): void => {
if (acc === "f64") {
B.line(`store double ${value}, ptr ${valuePtr}${host.fieldAliasAttachment(valuePtr)}`);
} else if (acc === "i32") {
B.line(`store i32 ${int32Value}, ptr ${valuePtr}${host.fieldAliasAttachment(valuePtr)}`);
} else if (acc === "bool") {
const bits = B.tmp();
B.line(`${bits} = zext i1 ${value} to i8`);
@@ -713,12 +781,21 @@ export function emitDenseArrayPop(
B.line(`${wide} = zext i8 ${raw} to i64`);
raw = wide;
} else if (slotTy === "i32") {
// The 8-byte slot of a compact reference is its address.
const wide = B.tmp(),
address = B.tmp();
B.line(`${wide} = zext i32 ${raw} to i64`);
B.line(`${address} = shl nuw nsw i64 ${wide}, 3`);
raw = address;
if (kind === "cref") {
// The 8-byte slot of a compact reference is its address.
const wide = B.tmp(),
address = B.tmp();
B.line(`${wide} = zext i32 ${raw} to i64`);
B.line(`${address} = shl nuw nsw i64 ${wide}, 3`);
raw = address;
} else {
// The slot bits of an int32 element's double, as the f64 path returns them.
const number = B.tmp(),
bits = B.tmp();
B.line(`${number} = sitofp i32 ${raw} to double`);
B.line(`${bits} = bitcast double ${number} to i64`);
raw = bits;
}
}
B.line(`store ${slotTy} 0, ptr ${valuePtr}${host.fieldAliasAttachment(valuePtr)}`);
// Only an array with states has a state byte to clear.
@@ -755,8 +832,8 @@ export function emitDenseArrayPush(
const B = host.B;
const size = host.sizeType;
const kind = element === null ? acc : slotKind(host, acc, element);
const ty = acc === "f64" ? "double" : acc === "bool" ? "i1" : "ptr";
const pusher = `scr_arr_push_${acc}`;
const ty = valueType(acc);
const pusher = `scr_arr_push_${runtimeAccess(acc)}`;
host.declare(`declare double @${pusher}(ptr, ${acc === "bool" ? "i1 zeroext" : ty})`);
const header = (field: number): string => {
const ptr = B.tmp();
@@ -794,6 +871,10 @@ export function emitDenseArrayPush(
const valuePtr = valuePointer(host, { offset: len, data }, kind);
if (acc === "f64") {
B.line(`store double ${value}, ptr ${valuePtr}${host.fieldAliasAttachment(valuePtr)}`);
} else if (acc === "i32") {
const int32 = B.tmp();
B.line(`${int32} = fptosi double ${value} to i32`);
B.line(`store i32 ${int32}, ptr ${valuePtr}${host.fieldAliasAttachment(valuePtr)}`);
} else if (acc === "bool") {
const bits = B.tmp();
B.line(`${bits} = zext i1 ${value} to i8`);
@@ -61,6 +61,7 @@ export function emitFunction(emitter: LlEmitter, fn: IrFunction): string {
const B = new BlockBuilder();
emitter.B = B;
emitter.currentFunctionName = fn.name;
emitter.findInt32Allocations(fn);
emitter.currentTier = fn.tier;
// Fiber switches restore the active exception cell before control
// returns to a synchronous frame (emitPendingCheck).
@@ -138,6 +139,8 @@ export function emitFunction(emitter: LlEmitter, fn: IrFunction): string {
emitter.speculativeFunction = fn.name;
emitter.speculation = null;
emitter.speculationResume = null;
emitter.int32Kinds = null;
emitter.int32KindResume = null;
emitter.finalFieldBorrows =
!fn.async &&
!fn.generator &&
@@ -34,9 +34,16 @@ import { dynHoleRead } from "./loop-array-borrows.js";
import { endsWithJump, matchStringSelfConcat } from "../../ir/analysis.js";
import { charAtUnitRead, emitDiscardedSplice } from "./expr-containers.js";
import { InternalCompilerError } from "../../errors.js";
import { emitDenseArrayGet, emitDenseArrayPop, emitDenseArraySet } from "./dense-array-access.js";
import {
emitDenseArrayGet,
emitDenseArrayPop,
emitDenseArraySet,
SCR_ELEM_I32,
} from "./dense-array-access.js";
import { emitBorrowedInput } from "./borrowed-inputs.js";
import { f64Lit, llvmCommentText, loadArrCount, sizeToDouble } from "./common.js";
import { f64Lit, llvmCommentText, loadArrCount, SCR_ARR_FIELD, sizeToDouble } from "./common.js";
import { planInt32KindLoop, type Int32KindLoopPlan } from "./int32-kind-loops.js";
import { withinInt32 } from "../../ir/integer-ranges.js";
import { resultSlotStore } from "../result-slots.js";
import {
matchIntegerArrayForLoop,
@@ -411,7 +418,21 @@ export function emitStmtBody(emitter: LlEmitter, s: IrStmt): void {
const v = emitter.emitExpr(s.value);
if (s.arr.type.kind !== "array")
throw new InternalCompilerError("llvm emitter bug: arraySet on non-array");
const acc = elemAccess(s.arr.type.elem);
if (emitter.checkedInt32Array(s.arr)) {
// An int32 array with checked stores: the int32 arm stores an int32
// value, and widens the array for any other first.
emitter.emitPublishedArrayGuard(arr.name, s.arr.type, 0, idx.name);
const elem = s.arr.type.elem;
const proven = emitter.provenDenseIndex(s.arr, s.index);
emitter.emitByElementKind(arr.name, null, (acc, widened) => {
if (acc === "i32") emitter.emitCheckedInt32Store(arr.name, v.name, s.value, widened);
emitDenseArraySet(emitter, arr.name, idx, s.index, acc, elem, v.name, proven);
return null;
});
break;
}
const acc = emitter.arrayAccess(s.arr, s.arr.type.elem);
const int32 = emitter.emitInt32KindStoreCheck(s.arr, v, s.value);
emitter.emitPublishedArrayGuard(arr.name, s.arr.type, 0, idx.name);
if (acc === "ref") emitter.toHeap(v);
emitDenseArraySet(
@@ -423,6 +444,7 @@ export function emitStmtBody(emitter: LlEmitter, s: IrStmt): void {
s.arr.type.elem,
v.name,
emitter.provenDenseIndex(s.arr, s.index),
int32,
);
break;
}
@@ -478,12 +500,19 @@ export function emitStmtBody(emitter: LlEmitter, s: IrStmt): void {
B.condBr(dropsLast, pop, call);
B.startBlock(pop);
const elem = s.arr.type.elem;
const acc = elemAccess(elem);
const raw = emitDenseArrayPop(emitter, arr.name, acc, elem, call, call);
if (acc === "ref" && isRefCounted(elem)) {
const old = B.tmp();
B.line(`${old} = inttoptr i64 ${raw} to ptr`);
emitter.releaseHeap(old, elem);
if (emitter.checkedInt32Array(s.arr)) {
emitter.emitByElementKind(arr.name, null, (acc) => {
emitDenseArrayPop(emitter, arr.name, acc, elem, call, call);
return null;
});
} else {
const acc = emitter.arrayAccess(s.arr, elem);
const raw = emitDenseArrayPop(emitter, arr.name, acc, elem, call, call);
if (acc === "ref" && isRefCounted(elem)) {
const old = B.tmp();
B.line(`${old} = inttoptr i64 ${raw} to ptr`);
emitter.releaseHeap(old, elem);
}
}
B.br(done);
B.startBlock(call);
@@ -1826,7 +1855,7 @@ export function emitTryCatch(emitter: LlEmitter, s: IrStmt & { kind: "tryCatch"
* case, refcounted/boxed case-body locals are NULL-reset up front and
* the scope-exit releases rely on NULL tolerance. */
export function emitSwitch(emitter: LlEmitter, s: IrStmt & { kind: "switch" }): void {
const unitSwitch = charAtUnitSwitch(s);
const unitSwitch = charAtUnitSwitch(emitter, s);
if (unitSwitch !== null) return emitSwitch(emitter, unitSwitch);
const B = emitter.B;
const discKind = s.disc.type.kind;
@@ -2098,9 +2127,26 @@ export function emitLoopCopies(
)
: null;
if (!plan) {
const proven = allowed ? planProvenLoop(emitter, loop) : null;
if (proven) emitProvenLoop(emitter, proven, exit, emitCopy);
else emitCopy(exit);
const ordinary = (exit: string): void => {
const proven = allowed ? planProvenLoop(emitter, loop) : null;
if (proven) emitProvenLoop(emitter, proven, exit, emitCopy);
else emitCopy(exit);
};
const kinds =
allowed &&
emitter.speculativeLoopsEnabled &&
emitter.int32Slots.checkedInt32Arrays.size > 0 &&
emitter.int32Kinds === null &&
emitter.speculation === null &&
emitter.speculationResume === null
? planInt32KindLoop(
loop,
(id) => emitter.checkedInt32Local(id),
(e) => withinInt32(emitter.integerRanges.get(e)),
)
: null;
if (kinds) emitInt32KindLoop(emitter, kinds, exit, kinds.writes.size > 0 ? emitCopy : ordinary);
else ordinary(exit);
return;
}
const B = emitter.B;
@@ -2243,6 +2289,124 @@ export function emitLoopCopies(
}
}
/** An int32 kind loop (int32-kind-loops.ts). The entry test reads the kind
* of each of the plan's arrays: when every one holds int32 slots the fast
* copy runs, when every one holds doubles the wide copy, which reads and
* writes them as double arrays, and otherwise (a plan with several arrays)
* the mixed copy, which tests their kinds at each access. A failed store of
* the fast copy re-executes in the third copy, which finishes its
* iteration with kind tests and then enters the mixed copy, or the wide
* copy when the plan has one array (it just widened). Each copy is emitted
* by `ordinary`: the loop's plain copy when a store may fail, since a
* proven integer loop keeps locals in shadows of its own copy that another
* copy cannot resume from. */
function emitInt32KindLoop(
emitter: LlEmitter,
plan: Int32KindLoopPlan,
exit: string,
ordinary: (exit: string) => void,
): void {
const B = emitter.B;
const fn = emitter.currentFunctionName;
const keys = new Set([...plan.arrays.keys()].map((id) => `${fn}\0${id}`));
const several = keys.size > 1;
let allInt32 = "true",
allWide = "true";
const and = (all: string, next: string): string => {
if (all === "true") return next;
const both = B.tmp();
B.line(`${both} = and i1 ${all}, ${next}`);
return both;
};
for (const ref of plan.arrays.values()) {
const arr = emitter.emitStableReceiver(ref, []);
const kindPtr = B.tmp(),
kind = B.tmp(),
isInt32 = B.tmp();
B.line(
`${kindPtr} = getelementptr inbounds %ScrArr, ptr ${arr.name}, i32 0, i32 ${SCR_ARR_FIELD.elem}`,
);
emitter.markMemoryPointer(kindPtr, "array:header");
B.line(`${kind} = load i8, ptr ${kindPtr}${emitter.fieldAliasAttachment(kindPtr)}`);
B.line(`${isInt32} = icmp eq i8 ${kind}, ${SCR_ELEM_I32}`);
allInt32 = and(allInt32, isInt32);
if (several) {
const isWide = B.tmp();
B.line(`${isWide} = xor i1 ${isInt32}, true`);
allWide = and(allWide, isWide);
}
}
const fast = B.newLabel("kind.fast"),
wide = B.newLabel("kind.wide"),
mixed = several ? B.newLabel("kind.mixed") : null;
if (mixed === null) B.condBr(allInt32, fast, wide);
else {
const other = B.newLabel("kind.other");
B.condBr(allInt32, fast, other);
B.startBlock(other);
B.condBr(allWide, wide, mixed);
}
const resume = new Map<IrStmt, { label: string; placed: boolean }>();
for (const s of plan.writes) resume.set(s, { label: B.newLabel("kind.resume"), placed: false });
// The header the third copy enters: the mixed copy's, or the wide copy's.
let header: string | null = null;
const capture = (label: string): void => {
header = label;
};
B.startBlock(wide);
if (resume.size > 0 && mixed === null) emitter.speculationHeader = capture;
emitter.int32Kinds = { int32: new Set(), wide: keys, resume: new Map() };
try {
ordinary(exit);
} finally {
emitter.speculationHeader = null;
emitter.int32Kinds = null;
}
if (mixed !== null) {
B.startBlock(mixed);
if (resume.size > 0) emitter.speculationHeader = capture;
try {
ordinary(exit);
} finally {
emitter.speculationHeader = null;
}
}
if (resume.size > 0) {
if (header === null)
throw new InternalCompilerError("llvm emitter bug: int32 kind loop copy without a header");
// A failed store finishes its iteration here, as in a speculative
// integer loop's third copy.
B.startBlock(B.newLabel("kind.tail"));
emitter.speculationResume = resume;
emitter.speculationHeader = () => {
B.br(header!);
B.startBlock(B.newLabel("kind.unused"));
};
try {
ordinary(exit);
} finally {
emitter.speculationResume = null;
emitter.speculationHeader = null;
}
}
B.startBlock(fast);
if (![...resume.values()].every((r) => r.placed)) {
// A write the third copy never reached: this copy tests the kinds.
ordinary(exit);
return;
}
emitter.int32Kinds = {
int32: keys,
wide: new Set(),
resume: new Map([...resume].map(([s, r]) => [s, r.label])),
};
try {
ordinary(exit);
} finally {
emitter.int32Kinds = null;
}
}
/** At the top of a loop copy's header: lets a speculative loop find its
* ordinary copy's header, and keeps its third copy out of the loop. */
function enterLoopHeader(emitter: LlEmitter, header: string): void {
@@ -2518,7 +2682,10 @@ function nullableStringSwitch(
* one-unit string literals, as a switch on `s.charCodeAt(i)` with the units
* as cases: no character string is built. An index out of range answers
* NaN, which matches no case, as "" matches no one-unit literal. */
function charAtUnitSwitch(s: IrStmt & { kind: "switch" }): (IrStmt & { kind: "switch" }) | null {
function charAtUnitSwitch(
emitter: LlEmitter,
s: IrStmt & { kind: "switch" },
): (IrStmt & { kind: "switch" }) | null {
const disc = s.disc;
if (disc.kind !== "strIntrinsic" || disc.method !== "charAt" || disc.args.length !== 1)
return null;
@@ -2531,7 +2698,7 @@ function charAtUnitSwitch(s: IrStmt & { kind: "switch" }): (IrStmt & { kind: "sw
return null;
return {
...s,
disc: charAtUnitRead(disc),
disc: charAtUnitRead(emitter, disc),
cases: s.cases.map((c) =>
c.test?.kind === "strLit"
? {
+299 -48
View File
@@ -79,8 +79,18 @@ import {
} from "./emit-journal.js";
import { specializeNumericCalls } from "../../ir/numeric-call-specialization.js";
import { versionLoopBoundParameters } from "../../ir/parameter-versions.js";
import {
allocatedLiteral,
fillAllocationLength,
selectInt32Arrays,
} from "../../ir/int32-arrays.js";
import { everyExpr, everyExprChild, everyStmtChild, everyStmtList } from "../../ir/traverse.js";
import { INT32_RANGE, type IntegerRange, type IntegerRanges } from "../../ir/integer-ranges.js";
import {
INT32_RANGE,
withinInt32,
type IntegerRange,
type IntegerRanges,
} from "../../ir/integer-ranges.js";
import { unshareNodes } from "../../ir/unshare.js";
import { copyStableCaptures } from "../../ir/capture-values.js";
import { resolveSelfCaptures } from "../../ir/self-captures.js";
@@ -100,7 +110,7 @@ import {
} from "./int32-abi.js";
import { type Int32Local } from "./int32-locals.js";
import { findIntegerClones, NO_INTEGER_CLONES, type IntegerClones } from "./integer-clones.js";
import { exactInteger, narrowInteger } from "./integer-values.js";
import { exactInteger, narrowInteger, widenInteger } from "./integer-values.js";
import { findInitializerBindings } from "../../ir/initializer-bindings.js";
import { findConstantNumericTables, type ConstantNumericTable } from "../../ir/constant-tables.js";
import {
@@ -147,7 +157,12 @@ import {
} from "./inline-unions.js";
import { OptionalArrayReads, type LocalArrayRead } from "./local-array-reads.js";
import { CheckedNarrows } from "./checked-narrows.js";
import { emitDenseArraySet, emitDenseSlotUnlink } from "./dense-array-access.js";
import {
emitDenseArraySet,
emitDenseSlotUnlink,
SCR_ELEM_I32,
type ArrayAccess,
} from "./dense-array-access.js";
import { matchMapRead, type MapReadLifetimes } from "./map-read-lifetimes.js";
import { type MapEntryFusion } from "./map-entry-fusion.js";
import { emitBorrowedFieldSequence, guardedValue } from "./borrowed-receivers.js";
@@ -168,7 +183,7 @@ import {
} from "./string-slices.js";
import { LazyCaptures } from "./lazy-captures.js";
import { StackCaptures } from "./stack-captures.js";
import { emitFieldAbsentTest, f64Lit } from "./common.js";
import { emitFieldAbsentTest, f64Lit, SCR_ARR_FIELD } from "./common.js";
import { ffiExtendsNarrowIntegers } from "../targets.js";
import {
emitLiteralExpr,
@@ -483,39 +498,65 @@ function versionParameters(mod: IrModule): IrModule {
return versionLoopBoundParameters(mod, mangleFunction);
}
export function emitLlvmModule(input: IrModule, options: LlvmTargetOptions = {}): string {
/** The int32 arrays the int32 slot analysis keeps in `mod` (proven and
* checked), under the inputs LlEmitter gives it (selectInt32Arrays). */
function provenInt32Arrays(mod: IrModule): ReadonlySet<string> {
const slots = analyzeInt32Slots(mod, {
dynamicStores: dynamicFieldStores(mod),
initializerBindings: findInitializerBindings(mod),
});
return new Set([...slots.int32Arrays, ...slots.checkedInt32Arrays]);
}
/** The emitter for `input` after the module passes every emission runs. */
function moduleEmitter(input: IrModule, options: LlvmTargetOptions): LlEmitter {
// Facts are keyed by node identity: emit the same module whether it was
// lowered in memory or read back from its serialized form.
const mod = unshareNodes(input);
// Keep source storage intact for debugger inspection in dev builds.
return new LlEmitter(
options.debugSources === undefined
? versionParameters(
addRecordResultEntries(
scalarizeNumericRecords(
specializeCallbackCalls(
specializeNumericCalls(
foldConstantGlobals(
copyStableCaptures(
resolveSelfCaptures(
hoistFinalFieldReads(
freezeInitOnlyGlobals(
holdCharLocalsAsUnits(dropUnwrittenPropertyBags(mod)),
? selectInt32Arrays(
versionParameters(
addRecordResultEntries(
scalarizeNumericRecords(
specializeCallbackCalls(
specializeNumericCalls(
foldConstantGlobals(
copyStableCaptures(
resolveSelfCaptures(
hoistFinalFieldReads(
freezeInitOnlyGlobals(
holdCharLocalsAsUnits(dropUnwrittenPropertyBags(mod)),
),
),
),
),
),
mangleFunction,
),
mangleFunction,
),
mangleFunction,
),
),
),
provenInt32Arrays,
)
: mod,
options,
).emit();
);
}
export function emitLlvmModule(input: IrModule, options: LlvmTargetOptions = {}): string {
return moduleEmitter(input, options).emit();
}
/** emitLlvmModule's text as parts that concatenate to it, without the copy
* that joining them makes: a large program's text is 150 MB. */
export function emitLlvmModuleParts(input: IrModule, options: LlvmTargetOptions = {}): string[] {
return moduleEmitter(input, options)
.emitParts()
.flatMap((part, index) => (index === 0 ? [part] : ["\n", part]));
}
/** Keep large translation units below the host's single-string limit. */
@@ -525,35 +566,7 @@ export function emitLlvmModuleSource(
): string | readonly string[] {
// IR validation, contracts and the frontend's disposal run after lowering without checks.
relieveHeapPressure(true);
const mod = unshareNodes(input);
const parts = new LlEmitter(
options.debugSources === undefined
? versionParameters(
addRecordResultEntries(
scalarizeNumericRecords(
specializeCallbackCalls(
specializeNumericCalls(
foldConstantGlobals(
copyStableCaptures(
resolveSelfCaptures(
hoistFinalFieldReads(
freezeInitOnlyGlobals(
holdCharLocalsAsUnits(dropUnwrittenPropertyBags(mod)),
),
),
),
),
),
mangleFunction,
),
mangleFunction,
),
),
),
)
: mod,
options,
).emitParts();
const parts = moduleEmitter(input, options).emitParts();
const length = parts.reduce((total, part) => total + part.length, Math.max(0, parts.length - 1));
if (length <= 256 * 1024 * 1024) return parts.join("\n");
return parts.flatMap((part, index) => (index === 0 ? [part] : ["\n", part]));
@@ -734,6 +747,228 @@ export class LlEmitter {
return `, !tbaa !${tag}`;
}
/** The current function's int32 array allocations: the array literals
* its int32 arrays are written with. They construct SCR_ELEM_I32 arrays. */
int32Allocations = new Set<IrExpr>();
/** The fill temporaries (`new Array(n).fill(v)`) of those allocations:
* the same arrays, so int32 too. */
int32ArrayTemps = new Set<string>();
/** The same for the current function's int32 arrays with checked stores,
* which are only allocated by fills (ir/int32-arrays.ts
* checkableInt32Allocations): their temporaries' literals, with the fill's
* length (a literal constructs SCR_ELEM_I32 storage when that length is at
* least CHECKED_INT32_MIN_LENGTH, SCR_ELEM_F64 otherwise), and their
* temporaries. */
checkedAllocations = new Map<IrExpr, IrExpr>();
checkedArrayTemps = new Set<string>();
/** The fill calls (`.fill(v)`) of those allocations: a value that is not
* an int32 widens the new array first. */
checkedFills = new Set<IrExpr>();
/** Collect the int32 allocations of `fn`, about to be emitted. */
findInt32Allocations(fn: IrFunction): void {
this.int32Allocations.clear();
this.int32ArrayTemps.clear();
this.checkedAllocations.clear();
this.checkedArrayTemps.clear();
this.checkedFills.clear();
const { int32Arrays, checkedInt32Arrays } = this.int32Slots;
if (int32Arrays.size === 0 && checkedInt32Arrays.size === 0) return;
everyStmtList(fn.body, {
stmt: (s) => {
if (s.kind !== "varDecl" && s.kind !== "assign") return true;
const key = `${fn.name}\0${s.localId}`;
const checked = checkedInt32Arrays.has(key);
if (!checked && !int32Arrays.has(key)) return true;
const value = s.kind === "varDecl" ? s.init : s.value;
const allocation = value && allocatedLiteral(value);
if (!allocation) return true;
const length = checked ? fillAllocationLength(value) : null;
if (length !== null) this.checkedAllocations.set(allocation.literal, length);
else if (!checked) this.int32Allocations.add(allocation.literal);
if (allocation.temporary !== null)
(checked ? this.checkedArrayTemps : this.int32ArrayTemps).add(allocation.temporary);
if (checked && value !== allocation.literal) this.checkedFills.add(value!);
return true;
},
expr: () => true,
});
}
/** Whether `arr` is one of the current function's int32 arrays
* (ir/int32-arrays.ts, kept by the int32 slot analysis). */
int32Array(arr: IrExpr): boolean {
if (arr.kind !== "varRef") return false;
if (this.int32KindArray(arr)) return true;
if (this.int32Slots.int32Arrays.size === 0) return false;
return (
this.int32ArrayTemps.has(arr.localId) ||
this.int32Slots.int32Arrays.has(`${this.currentFunctionName}\0${arr.localId}`)
);
}
/** Whether `arr` is one of the current function's int32 arrays with
* checked stores: int32 slots until a store widens it to doubles. Every
* access tests its kind (emitByElementKind), except in the fast copy of
* an int32 kind loop, which reads and writes it as an int32 array. */
checkedInt32Array(arr: IrExpr): boolean {
if (arr.kind !== "varRef" || this.int32KindArray(arr)) return false;
if (this.int32Kinds?.wide.has(`${this.currentFunctionName}\0${arr.localId}`)) return false;
return this.checkedInt32Local(arr.localId);
}
/** Whether the local `localId` holds one of the current function's int32
* arrays with checked stores. */
checkedInt32Local(localId: string): boolean {
if (this.int32Slots.checkedInt32Arrays.size === 0) return false;
return (
this.checkedArrayTemps.has(localId) ||
this.int32Slots.checkedInt32Arrays.has(`${this.currentFunctionName}\0${localId}`)
);
}
/** Whether `arr` is an int32 array with checked stores that the fast copy
* of the int32 kind loop being emitted holds as an int32 array. */
int32KindArray(arr: IrExpr): boolean {
return (
this.int32Kinds !== null &&
arr.kind === "varRef" &&
this.int32Kinds.int32.has(`${this.currentFunctionName}\0${arr.localId}`)
);
}
/** Before a store of `value` (of `expr`) into `arr`, in the fast copy of
* an int32 kind loop: continue when it is an int32 that is not -0, and
* otherwise leave for the statement's resume point, where it re-executes
* with kind tests. Returns the value as an i32 when it is at hand as an
* integer, or null when the store converts the double itself. */
emitInt32KindStoreCheck(arr: IrExpr, value: LlValue, expr: IrExpr): string | null {
if (!this.int32KindArray(arr) || withinInt32(this.integerRanges.get(expr))) return null;
const integer = exactInteger(this, value, expr);
if (integer !== null && withinInt32(integer.range)) return narrowInteger(this, integer);
const resume = this.int32KindResume;
if (resume === null)
throw new InternalCompilerError("llvm emitter bug: an int32 kind store without a resume");
const B = this.B;
const ok = B.tmp(),
int = B.tmp();
if (integer !== null) {
// An exact integer: a range test on its 64-bit form.
const wide = widenInteger(this, integer),
low = B.tmp(),
high = B.tmp();
B.line(`${low} = icmp sge i64 ${wide}, -2147483648`);
B.line(`${high} = icmp sle i64 ${wide}, 2147483647`);
B.line(`${ok} = and i1 ${low}, ${high}`);
const store = B.newLabel("arr.store.i32");
B.condBr(ok, store, resume);
B.startBlock(store);
B.line(`${int} = trunc i64 ${wide} to i32`);
return int;
}
this.declare("declare i32 @llvm.fptosi.sat.i32.f64(double)");
const back = B.tmp(),
exact = B.tmp(),
bits = B.tmp(),
notNegativeZero = B.tmp();
B.line(`${int} = call i32 @llvm.fptosi.sat.i32.f64(double ${value.name})`);
B.line(`${back} = sitofp i32 ${int} to double`);
B.line(`${exact} = fcmp oeq double ${back}, ${value.name}`);
B.line(`${bits} = bitcast double ${value.name} to i64`);
B.line(`${notNegativeZero} = icmp ne i64 ${bits}, -9223372036854775808`);
B.line(`${ok} = and i1 ${exact}, ${notNegativeZero}`);
const store = B.newLabel("arr.store.i32");
B.condBr(ok, store, resume);
B.startBlock(store);
return int;
}
/** Emit an access to the checked int32 array `array` once per element
* kind: `arm("i32", widened)` while it holds int32 slots and
* `arm("f64", widened)` after it widened, joined after both. `widened` is
* the label of the double arm, where an int32 arm's store of a value that
* is not an int32 continues after widening the array. Returns the phi of
* the arms' results (of LLVM type `ty`), or null when `ty` is null. */
emitByElementKind(
array: string,
ty: string | null,
arm: (acc: "i32" | "f64", widened: string) => string | null,
): string | null {
const B = this.B;
const narrow = B.newLabel("arr.kind.i32"),
wide = B.newLabel("arr.kind.f64"),
join = B.newLabel("arr.kind.join");
const kindPtr = B.tmp(),
kind = B.tmp(),
isInt32 = B.tmp();
B.line(
`${kindPtr} = getelementptr inbounds %ScrArr, ptr ${array}, i32 0, i32 ${SCR_ARR_FIELD.elem}`,
);
this.markMemoryPointer(kindPtr, "array:header");
B.line(`${kind} = load i8, ptr ${kindPtr}${this.fieldAliasAttachment(kindPtr)}`);
B.line(`${isInt32} = icmp eq i8 ${kind}, ${SCR_ELEM_I32}`);
B.condBr(isInt32, narrow, wide);
B.startBlock(narrow);
const narrowValue = arm("i32", wide);
const narrowEnd = B.isTerminated() ? null : B.currentLabel();
if (narrowEnd !== null) B.br(join);
B.startBlock(wide);
const wideValue = arm("f64", wide);
const wideEnd = B.isTerminated() ? null : B.currentLabel();
if (wideEnd !== null) B.br(join);
B.startBlock(join);
// Arms that end in their own branches (a read's shared join) leave this
// join without predecessors.
if (narrowEnd === null && wideEnd === null) B.terminate("unreachable");
if (ty === null) return null;
const arms: string[] = [];
if (narrowEnd !== null) arms.push(`[ ${narrowValue}, %${narrowEnd} ]`);
if (wideEnd !== null) arms.push(`[ ${wideValue}, %${wideEnd} ]`);
const result = B.tmp();
B.line(`${result} = phi ${ty} ${arms.join(", ")}`);
return result;
}
/** Before storing the double `value` (of `expr`) into an int32 arm
* (emitByElementKind) of the checked int32 array `array`: continue in this
* arm when it is an int32 that is not -0, or widen the array and continue
* at `widened`. A value proven an int32 needs no test. */
emitCheckedInt32Store(array: string, value: string, expr: IrExpr, widened: string): void {
if (withinInt32(this.integerRanges.get(expr))) return;
const B = this.B;
this.declare("declare i32 @llvm.fptosi.sat.i32.f64(double)");
this.declare("declare void @scr_arr_widen_i32(ptr)");
const int = B.tmp(),
back = B.tmp(),
exact = B.tmp(),
bits = B.tmp(),
notNegativeZero = B.tmp(),
ok = B.tmp();
B.line(`${int} = call i32 @llvm.fptosi.sat.i32.f64(double ${value})`);
B.line(`${back} = sitofp i32 ${int} to double`);
B.line(`${exact} = fcmp oeq double ${back}, ${value}`);
B.line(`${bits} = bitcast double ${value} to i64`);
B.line(`${notNegativeZero} = icmp ne i64 ${bits}, -9223372036854775808`);
B.line(`${ok} = and i1 ${exact}, ${notNegativeZero}`);
const store = B.newLabel("arr.store.i32"),
widen = B.newLabel("arr.widen");
B.condBr(ok, store, widen);
B.startBlock(widen);
B.line(`call void @scr_arr_widen_i32(ptr ${array})`);
B.br(widened);
B.startBlock(store);
}
/** The access an element operation on `arr` (of element type `elem`)
* takes: int32 arrays read and write their 4-byte slots. */
arrayAccess(arr: IrExpr, elem: IrType): ArrayAccess {
if (this.checkedInt32Array(arr))
throw new InternalCompilerError(
"llvm emitter bug: a checked int32 array without its kind test",
);
return this.int32Array(arr) ? "i32" : elemAccess(elem);
}
/** Whether an element access at `index` of `arr` lies in the hole-free
* prefix: the counter of a dense array loop's fast copy indexing one of
* the arrays its entry test covered. */
@@ -1180,6 +1415,14 @@ export class LlEmitter {
* (`enterLoopHeader`): records the ordinary copy's header, or sends the
* third copy's header there. */
speculationHeader: ((header: string) => void) | null = null;
/** A copy of an int32 kind loop (int32-kind-loops.ts) being emitted: the
* int32 arrays with checked stores it reads and writes as int32 arrays
* (the fast copy) or as double arrays (an ordinary copy entered only
* after they widened), as `function\0local` keys, and where each of the
* fast copy's writes resumes in the third copy. */
int32Kinds: { int32: Set<string>; wide: Set<string>; resume: Map<IrStmt, string> } | null = null;
/** The resume label of the fast-copy write being emitted. */
int32KindResume: string | null = null;
/** The function being emitted, for speculative globals. */
speculativeFunction = "";
/** Every function that writes each number global (computed on demand). */
@@ -1201,6 +1444,10 @@ export class LlEmitter {
* (findIntegerOrNaNLocals in wrapping-sums.ts). */
integerOrNaNLocals: ReadonlyMap<string, IntegerRange> = new Map();
bytesBounds: ReadonlySet<IrExpr | IrStmt> = new Set();
/** The charCodeAt reads standing for charAt reads the emitter rewrote
* (one-unit tests and switches, charAtUnitRead), with the charAt each
* stands for: a loop version's proof for the charAt covers its read. */
readonly charAtUnitReads = new Map<IrExpr, IrExpr>();
/** Worker programs: typed-array accesses whose views an enclosing version
* guard proved thread-owned. */
bytesOwned: ReadonlySet<IrExpr | IrStmt> = new Set();
@@ -5269,10 +5516,14 @@ export class LlEmitter {
const loop =
s.kind === "while" || s.kind === "doWhile" || s.kind === "for" || s.kind === "forOf";
if (loop) this.loopDepth++;
// An int32 kind loop's fast-copy write leaves for its resume point.
const kindResume = this.int32KindResume;
this.int32KindResume = this.int32Kinds?.resume.get(s) ?? null;
try {
this.emitStmtBody(s);
} finally {
this.B.debugLocation = previous;
this.int32KindResume = kindResume;
if (loop) this.loopDepth--;
}
}
@@ -6519,7 +6770,7 @@ export class LlEmitter {
return tagInSet(this, uName, unionId, tags);
}
arrPush(arr: string, acc: "f64" | "bool" | "ref", value: string, element: IrType | null): string {
arrPush(arr: string, acc: ArrayAccess, value: string, element: IrType | null): string {
return arrPush(this, arr, acc, value, element);
}
@@ -1118,7 +1118,9 @@ export function emitInlineAtomicsOp(
if (op !== 0 && op !== 1 && op !== 3 && ATOMIC_RMW[op] === undefined) return null;
const elem = array.type.elem;
const B = host.B;
const receiver = host.emitExpr(array).name;
// The instruction and scr_atomics_op borrow the array: an unchanged local
// owns it across the later operands and the operation.
const receiver = emitBorrowedInput(host, array).name;
const idxValue = host.emitExpr(index);
const operand = host.emitExpr(value);
const swap = host.emitExpr(replacement);
@@ -60,7 +60,10 @@ import {
emitDenseArrayNumber,
emitDenseArrayPop,
emitDenseArrayPush,
runtimeAccess,
type ArrayAccess,
} from "./dense-array-access.js";
import { INT32_ARRAY_METHODS } from "../../ir/int32-arrays.js";
import { emitUnionWiden } from "./expr-records.js";
import { registerReplay } from "./emit-journal.js";
@@ -115,7 +118,7 @@ export function tagInSet(
export function arrPush(
host: LlvmEmitterContext,
arr: string,
acc: "f64" | "bool" | "ref",
acc: ArrayAccess,
value: string,
element: IrType | null,
): string {
@@ -126,7 +129,7 @@ export function arrPush(
export function emitArrayValues(
host: LlvmEmitterContext,
arr: string,
acc: "f64" | "bool" | "ref",
acc: ArrayAccess,
element: IrType,
values: LlValue[],
prepend = false,
@@ -142,12 +145,15 @@ export function emitArrayValues(
}
if (values.length === 1) {
if (!prepend) return arrPush(host, arr, acc, values[0]!.name, element);
const ty = acc === "f64" ? "double" : acc === "bool" ? "i1" : "ptr";
const flat = runtimeAccess(acc);
const ty = flat === "f64" ? "double" : flat === "bool" ? "i1" : "ptr";
host.declare(
`declare double @scr_arr_unshift_${acc}(ptr, ${ty}${acc === "bool" ? " zeroext" : ""})`,
`declare double @scr_arr_unshift_${flat}(ptr, ${ty}${flat === "bool" ? " zeroext" : ""})`,
);
const result = B.tmp();
B.line(`${result} = call double @scr_arr_unshift_${acc}(ptr ${arr}, ${ty} ${values[0]!.name})`);
B.line(
`${result} = call double @scr_arr_unshift_${flat}(ptr ${arr}, ${ty} ${values[0]!.name})`,
);
return result;
}
const capacity = Math.min(values.length, 64);
@@ -165,8 +171,9 @@ export function emitArrayValues(
const slot = B.tmp(),
pointer = B.tmp();
const value = values[start + i]!.name;
// The runtime takes double bits for int32 arrays too, and converts.
const pack =
acc === "f64"
acc === "f64" || acc === "i32"
? `bitcast double ${value} to i64`
: acc === "bool"
? `zext i1 ${value} to i64`
@@ -286,14 +293,11 @@ export function emitArrayCopyLoop(
host.B.line(`call double @scr_arr_push_spread(ptr ${dst}, ptr ${src})`);
}
/** The `charCodeAt` reads that stand for `charAt` reads the emitter
* rewrote, such as one-unit tests and switches: a loop version's proof
* for the charAt covers its read. */
const unitReadOrigins = new WeakMap<IrExpr, IrExpr>();
/** `s.charCodeAt(i)` in place of `s.charAt(i)`, the same index read as a
* code unit. */
* code unit. The emitter records the charAt it stands for
* (charAtUnitReads), so a loop version's proof for the charAt covers it. */
export function charAtUnitRead(
host: LlvmEmitterContext,
charAt: IrExpr & { kind: "strIntrinsic" },
): IrExpr & { kind: "strIntrinsic" } {
const read: IrExpr & { kind: "strIntrinsic" } = {
@@ -304,7 +308,7 @@ export function charAtUnitRead(
type: F64,
loc: charAt.loc,
};
unitReadOrigins.set(read, charAt);
host.charAtUnitReads.set(read, charAt);
return read;
}
@@ -312,7 +316,7 @@ export function charAtUnitRead(
* below the length (bytes-loop-versions.ts). */
function provenStringRead(host: LlvmEmitterContext, e: IrExpr): boolean {
if (host.bytesBounds.has(e)) return true;
const origin = unitReadOrigins.get(e);
const origin = host.charAtUnitReads.get(e);
return origin !== undefined && host.bytesBounds.has(origin);
}
@@ -610,7 +614,15 @@ export function emitArrIntrinsic(
if (e.receiver.type.kind !== "array")
throw new InternalCompilerError("llvm emitter bug: arrIntrinsic on non-array");
const elem = e.receiver.type.elem;
const acc = elemAccess(elem);
// Int32 arrays (ir/int32-arrays.ts) reach only these methods; the others
// take the runtime family's flat access. One with checked stores tests
// its kind at each inline access (emitByElementKind) and otherwise takes
// the double entry points, which serve both kinds.
const checked = host.checkedInt32Array(e.receiver);
const access = checked ? "f64" : host.arrayAccess(e.receiver, elem);
if ((access === "i32" || checked) && !INT32_ARRAY_METHODS.has(e.method))
throw new InternalCompilerError(`llvm emitter bug: ${e.method} on an int32 array`);
const acc = runtimeAccess(access);
const accTy = acc === "f64" ? "double" : acc === "bool" ? "i1" : "ptr";
const accArg = acc === "bool" ? "i1 zeroext" : accTy;
const method = e.method;
@@ -631,6 +643,13 @@ export function emitArrIntrinsic(
e.receiver.kind === "varRef"
? host.constantNumericTables.get(e.receiver.localId)
: undefined;
if (!table && elem.kind === "f64" && checked) {
const proven = host.provenDenseIndex(e.receiver, e.args[0]!);
const value = host.emitByElementKind(r.name, "double", (kind) =>
emitDenseArrayNumber(host, r.name, index, e.args[0], proven, kind),
)!;
return { name: value, type: e.type };
}
if (!table && elem.kind === "f64")
return {
name: emitDenseArrayNumber(
@@ -639,6 +658,7 @@ export function emitArrIntrinsic(
index,
e.args[0],
host.provenDenseIndex(e.receiver, e.args[0]!),
access === "i32" ? "i32" : "f64",
),
type: e.type,
};
@@ -675,7 +695,22 @@ export function emitArrIntrinsic(
// no-op zero-argument call.
const vs = e.args.map((a) => host.emitExpr(a));
host.emitPublishedArrayGuard(r.name, e.receiver.type, 1, f64Lit(0), f64Lit(vs.length));
return { name: emitArrayValues(host, r.name, acc, elem, vs), type: e.type };
if (checked && vs.length > 0) {
// Each value appends in its own kind test: one that is not an int32
// widens the array before it is pushed.
let length = "";
vs.forEach((v, i) => {
length = host.emitByElementKind(r.name, "double", (kind, widened) => {
if (kind === "i32") host.emitCheckedInt32Store(r.name, v.name, e.args[i]!, widened);
return arrPush(host, r.name, kind, v.name, elem);
})!;
});
return { name: length, type: e.type };
}
// In an int32 kind loop's fast copy, every value is checked before
// the first appends.
vs.forEach((v, i) => host.emitInt32KindStoreCheck(e.receiver, v, e.args[i]!));
return { name: emitArrayValues(host, r.name, access, elem, vs), type: e.type };
}
case "pushSpread": {
// `a.push(...src)`: append src's elements in order (borrowed src,
@@ -749,7 +784,13 @@ export function emitArrIntrinsic(
let slowLabel = "";
if (method === "pop") {
slowLabel = B.newLabel("remove.slow");
denseRaw = emitDenseArrayPop(host, r.name, acc, elem, la, slowLabel);
const empty = la,
slow = slowLabel;
denseRaw = checked
? host.emitByElementKind(r.name, "i64", (kind) =>
emitDenseArrayPop(host, r.name, kind, elem, empty, slow),
)!
: emitDenseArrayPop(host, r.name, access, elem, la, slowLabel);
denseLabel = B.currentLabel();
B.br(lp);
B.startBlock(slowLabel);
@@ -938,6 +979,14 @@ export function emitArrIntrinsic(
case "fillUndefined": {
const args = e.args.map((arg) => host.emitExpr(arg));
host.emitPublishedArrayGuard(r.name, e.receiver.type, 7);
if (host.checkedFills.has(e)) {
// A new int32 array with checked stores, filled with a value that
// may not be an int32: widen it first then.
const filled = B.newLabel("arr.fill.kind");
host.emitCheckedInt32Store(r.name, args[0]!.name, e.args[0]!, filled);
B.br(filled);
B.startBlock(filled);
}
const helper =
e.method === "copyWithin"
? "scr_arr_copy_within"
@@ -15,11 +15,23 @@ export interface LlValue {
/** An exact integer snapshot, excluding negative zero, with no discarded
* high bits and the proven range of its original value. */
integer?: { name: string; type: "i32" | "i64"; signed: boolean; range: IntegerRange };
/** An i64 holding the exact integer that the arithmetic producing this
* number computes, when that integer has no proven safe range: its
* absolute value is at most `bound`, and the double equals it whenever
* its absolute value is below `below` (a zero may be -0). Index consumers
* use it in place of a round trip (integer-values.ts `indexInteger`). */
wideInteger?: WideInteger;
slot?: boolean;
/** A sequence local held in a capture box; release the box itself. */
boxed?: boolean;
}
export interface WideInteger {
name: string;
below: number;
bound: number;
}
export type ExprOf<K extends IrExpr["kind"]> = Extract<IrExpr, { kind: K }>;
export type LibCallExpr = ExprOf<"libCall">;
type LibCallPrefixOf<T extends string> = T extends `${infer Prefix}.${string}` ? Prefix : never;
@@ -8,6 +8,7 @@ import { LlvmUnsupportedError } from "./unsupported.js";
import type { LlvmEmitterContext, ExprOf, LlValue } from "./expr-context.js";
import { BYTES_ELEM_NUM, closureIdentityEqual, f64Lit } from "./common.js";
import { CHECKED_INT32_MIN_LENGTH } from "../../ir/int32-arrays.js";
import { emitStringInputs } from "./string-lifetimes.js";
import { emitStringParts, numberPart, stringParts } from "./string-construction.js";
import { emitBorrowedInputs } from "./borrowed-inputs.js";
@@ -16,10 +17,12 @@ import {
int32Number,
integerNumber,
narrowInteger,
wideArithmetic,
widenInteger,
} from "./integer-values.js";
import {
bitwiseIdentityOperand,
INT32_RANGE,
integerArithmeticRange,
isBitwiseOp,
withinInt32,
@@ -32,7 +35,15 @@ import {
emitPowerOfTwoRemainder,
emitSignedIntegerRemainder,
} from "./integer-remainder.js";
import { denseSlotWidth, emitDenseArrayGet, emitDenseArrayState } from "./dense-array-access.js";
import {
denseSlotWidth,
emitDenseArrayRead,
emitDenseArrayState,
runtimeAccess,
SCR_ELEM_I32,
slotWidth,
type ArrayAccess,
} from "./dense-array-access.js";
import { emitWrappingSum, markWrappingOperands } from "./wrapping-sums.js";
export function emitLiteralExpr(
@@ -337,7 +348,14 @@ export function emitOperatorExpr(
if (arith[e.op] !== undefined) {
B.line(`${t} = ${arith[e.op]} double ${l.name}, ${r.name}`);
const uint32 = emitWrappingSum(host, e, l, r, li, ri);
if (uint32 !== undefined) return { name: t, type: e.type, uint32 };
// The exact integer beside the double, for index consumers.
const wideInteger = wideArithmetic(host, e.op, l, li, r, ri);
if (uint32 !== undefined || wideInteger) {
const value: LlValue = { name: t, type: e.type };
if (uint32 !== undefined) value.uint32 = uint32;
if (wideInteger) value.wideInteger = wideInteger;
return value;
}
} else B.line(`${t} = fcmp ${cmp[e.op]} double ${l.name}, ${r.name}`);
} else if (bit[e.op] !== undefined) {
if (e.left.type.kind !== "f64" || e.right.type.kind !== "f64") {
@@ -648,7 +666,7 @@ export function emitStringExpr(
// equals no unit, as "" equals no one-unit literal.
const unit = charAtUnitTest(e);
if (unit !== null) {
const code = host.emitExpr(charAtUnitRead(unit.charAt));
const code = host.emitExpr(charAtUnitRead(host, unit.charAt));
const t = B.tmp();
B.line(
`${t} = fcmp ${e.negated ? "une" : "oeq"} double ${code.name}, ${f64Lit(unit.code)}`,
@@ -682,7 +700,7 @@ export function emitStringExpr(
// the code unit, as the equality above does.
const unit = charAtUnitTest(e);
if (unit !== null) {
const code = host.emitExpr(charAtUnitRead(unit.charAt));
const code = host.emitExpr(charAtUnitRead(host, unit.charAt));
const t = B.tmp();
B.line(
`${t} = fcmp ${charAtUnitPredicate(e.op, unit.left)} double ${code.name}, ${f64Lit(unit.code)}`,
@@ -886,9 +904,46 @@ export function emitContainerExpr(
// the plain elements up to the next one (scr_arr_reserve_spread).
const firstSpread = e.elems.findIndex((_, i) => spreadSet.has(i));
const capacity = small ? 0 : (e.capacity ?? (firstSpread < 0 ? e.elems.length : firstSpread));
B.line(`${arr} = ${arrNewCall(host.shapeHost, elem, String(capacity), small)}`);
// An int32 array's allocation (ir/int32-arrays.ts) constructs
// SCR_ELEM_I32 storage; its elements are proven int32s. The empty
// temporary of a fill that allocates an int32 array with checked
// stores takes SCR_ELEM_I32 storage when the fill's length (which
// evaluates without effects) is at least CHECKED_INT32_MIN_LENGTH, and
// SCR_ELEM_F64 storage otherwise.
const fillLength = host.checkedAllocations.get(e);
if (fillLength !== undefined && e.elems.length === 0) {
const length = host.emitExpr(fillLength);
const long = B.tmp(),
kind = B.tmp();
B.line(`${long} = fcmp oge double ${length.name}, ${f64Lit(CHECKED_INT32_MIN_LENGTH)}`);
B.line(`${kind} = select i1 ${long}, i32 ${SCR_ELEM_I32}, i32 0`);
host.declare(
small
? `declare ptr @scr_arr_new_small(i32)`
: `declare ptr @scr_arr_new(i32, ${host.sizeType})`,
);
B.line(
small
? `${arr} = call ptr @scr_arr_new_small(i32 ${kind})`
: `${arr} = call ptr @scr_arr_new(i32 ${kind}, ${host.sizeType} ${capacity})`,
);
return host.own({ name: arr, type: e.type });
}
const int32 = host.int32Allocations.has(e);
if (int32) {
host.declare(
small
? `declare ptr @scr_arr_new_small(i32)`
: `declare ptr @scr_arr_new(i32, ${host.sizeType})`,
);
B.line(
small
? `${arr} = call ptr @scr_arr_new_small(i32 ${SCR_ELEM_I32})`
: `${arr} = call ptr @scr_arr_new(i32 ${SCR_ELEM_I32}, ${host.sizeType} ${capacity})`,
);
} else B.line(`${arr} = ${arrNewCall(host.shapeHost, elem, String(capacity), small)}`);
const out = host.own({ name: arr, type: e.type });
const acc = elemAccess(elem);
const acc: ArrayAccess = int32 ? "i32" : elemAccess(elem);
let pending: LlValue[] = [];
let buffer: string | undefined;
const flush = (): void => {
@@ -909,7 +964,8 @@ export function emitContainerExpr(
B.line(
`call void @scr_arr_reserve_spread(ptr ${arr}, ptr ${v.name}, ${host.sizeType} ${plain})`,
);
host.emitArrayCopyLoop(arr, v.name, acc);
// A literal with spreads is never an int32 allocation.
host.emitArrayCopyLoop(arr, v.name, runtimeAccess(acc));
return;
}
pending.push(host.emitExpr(el));
@@ -954,33 +1010,65 @@ export function emitContainerExpr(
// Ref-element reads return +1 (the inline dense path retains like
// the runtime getter); own registers the owned temp in the frame.
const elem = e.arr.type.elem;
const acc = elemAccess(elem);
const t = emitDenseArrayGet(
if (host.checkedInt32Array(e.arr)) {
const proven = host.provenDenseIndex(e.arr, e.index);
const value = host.emitByElementKind(
arr.name,
"double",
(kind) =>
emitDenseArrayRead(
host,
arr.name,
idx,
e.index,
kind,
elem,
"scr_arr_get_f64",
true,
proven,
).value,
)!;
return { name: value, type: e.type };
}
const acc = host.arrayAccess(e.arr, elem);
const read = emitDenseArrayRead(
host,
arr.name,
idx,
e.index,
acc,
elem,
`scr_arr_get_${acc}`,
`scr_arr_get_${runtimeAccess(acc)}`,
true,
host.provenDenseIndex(e.arr, e.index),
);
return host.own({ name: t, type: e.type });
if (read.int32 !== null)
// An int32 array's element: its i32 is exact for index consumers.
return {
name: read.value,
type: e.type,
uint32: read.int32,
integer: { name: read.int32, type: "i32", signed: true, range: INT32_RANGE },
};
return host.own({ name: read.value, type: e.type });
}
case "arrayHas": {
const arr = host.emitStableReceiver(e.arr, [e.index]);
const idx = host.emitExpr(e.index);
if (e.arr.type.kind !== "array")
throw new InternalCompilerError("llvm emitter bug: arrayHas on non-array");
const t = emitDenseArrayState(
host,
arr.name,
idx,
e.index,
"has",
denseSlotWidth(host, e.arr.type.elem),
);
const t = host.checkedInt32Array(e.arr)
? host.emitByElementKind(arr.name, "i1", (kind) =>
emitDenseArrayState(host, arr.name, idx, e.index, "has", slotWidth(kind)),
)!
: emitDenseArrayState(
host,
arr.name,
idx,
e.index,
"has",
host.int32Array(e.arr) ? slotWidth("i32") : denseSlotWidth(host, e.arr.type.elem),
);
return { name: t, type: e.type };
}
case "arrayState": {
@@ -988,14 +1076,18 @@ export function emitContainerExpr(
const idx = host.emitExpr(e.index);
if (e.arr.type.kind !== "array")
throw new InternalCompilerError("llvm emitter bug: arrayState on non-array");
const t = emitDenseArrayState(
host,
arr.name,
idx,
e.index,
"state",
denseSlotWidth(host, e.arr.type.elem),
);
const t = host.checkedInt32Array(e.arr)
? host.emitByElementKind(arr.name, "double", (kind) =>
emitDenseArrayState(host, arr.name, idx, e.index, "state", slotWidth(kind)),
)!
: emitDenseArrayState(
host,
arr.name,
idx,
e.index,
"state",
host.int32Array(e.arr) ? slotWidth("i32") : denseSlotWidth(host, e.arr.type.elem),
);
return { name: t, type: e.type };
}
case "arrIntrinsic":
@@ -0,0 +1,135 @@
import type { IrExpr, IrStmt } from "../../ir/ir.js";
import { everyExprChild, everyStmtChild } from "../../ir/traverse.js";
import { MAX_LOOP_NODES, SIMPLE_STATEMENTS } from "./speculative-integers.js";
/** Int32 kind loops.
*
* An int32 array with checked stores (ir/int32-slots.ts) holds int32 slots
* (or doubles, when its fill was short) until a store of a value that is
* not an int32 widens it to doubles in place, so an access to it tests its
* element kind first (LlEmitter.emitByElementKind). An innermost loop that
* accesses such arrays tests their kinds once on entry instead
* (emitLoopCopies). The fast copy runs when every one of them holds int32
* slots and reads and writes them as int32 arrays; the wide copy runs when
* every one holds doubles and accesses them as double arrays; a loop over
* several arrays whose kinds differ runs the mixed copy, which tests the
* kinds at each access. A store of the fast copy whose value is not an
* int32 branches to the same statement in a third copy, which re-executes
* it with kind tests (widening the array) and finishes the iteration, then
* enters the mixed copy, or the wide copy when the loop has one array.
* Re-execution is exact because such a store's statement evaluates nothing
* but locals, literals, number arithmetic and element reads before it
* writes.
*
* Nothing else can widen these arrays: they never reach code other than
* the compiled element operations of their own function
* (ir/int32-arrays.ts). */
export interface Int32KindLoopPlan {
/** The arrays the fast copy reads and writes as int32 arrays, by local,
* with one of their references in the loop. */
arrays: Map<string, IrExpr>;
/** Element writes and pushes whose value may not be an int32: the
* statements the fast copy may leave at. */
writes: Set<IrStmt>;
}
type KindLoop = Extract<IrStmt, { kind: "for" | "while" | "doWhile" }>;
/** A number or boolean expression whose evaluation has no effect: locals,
* literals, arithmetic, comparisons and element reads of local arrays. */
function isPure(e: IrExpr): boolean {
if (e.type.kind !== "f64" && e.type.kind !== "bool") return false;
switch (e.kind) {
case "numLit":
case "boolLit":
case "varRef":
return true;
case "bin":
return (
(e.left.type.kind === "f64" || e.left.type.kind === "bool") &&
isPure(e.left) &&
isPure(e.right)
);
case "unary":
return isPure(e.operand);
case "ternary":
return isPure(e.cond) && isPure(e.then) && isPure(e.else_);
case "arrayGet":
case "arrayHas":
return e.arr.kind === "varRef" && isPure(e.index);
default:
return false;
}
}
/** The plan for `loop`, or null. `checked` names the int32 array locals
* with checked stores of the function; `provenInt32` the values the static
* range analysis proves int32s (stores that never leave the fast copy). */
export function planInt32KindLoop(
loop: KindLoop,
checked: (localId: string) => boolean,
provenInt32: (e: IrExpr) => boolean,
): Int32KindLoopPlan | null {
const arrays = new Map<string, IrExpr>();
const writes = new Set<IrStmt>();
const assigned = new Set<string>();
let nodes = 0;
let rejected = false;
const reject = (): false => {
rejected = true;
return false;
};
const array = (e: IrExpr): string | null =>
e.kind === "varRef" && checked(e.localId) ? e.localId : null;
const push = (e: IrExpr): (IrExpr & { kind: "arrIntrinsic" }) | null =>
e.kind === "arrIntrinsic" && e.method === "push" && array(e.receiver) !== null ? e : null;
const expr = (e: IrExpr): boolean => {
if (rejected) return false;
nodes++;
if (e.kind === "varRef" && checked(e.localId) && !arrays.has(e.localId))
arrays.set(e.localId, e);
// A push the statement walk did not take as a write: its values must
// be proven int32s, since it cannot leave the fast copy.
const pushed = push(e);
if (pushed && !pushed.args.every(provenInt32)) return reject();
return everyExprChild(e, expr, nested);
};
// Statements inside expressions run mid-expression, where the fast copy
// cannot leave: they may not write these arrays.
const nested = (s: IrStmt): boolean => {
nodes++;
if (s.kind === "assign" || s.kind === "varDecl") assigned.add(s.localId);
if (s.kind === "arraySet" && array(s.arr) !== null && !provenInt32(s.value)) return reject();
return everyStmtChild(s, expr, nested);
};
const stmt = (s: IrStmt): boolean => {
if (rejected) return false;
nodes++;
if (!SIMPLE_STATEMENTS.has(s.kind)) return reject();
if (s.kind === "assign" || s.kind === "varDecl") assigned.add(s.localId);
if (s.kind === "arraySet" && array(s.arr) !== null && !provenInt32(s.value)) {
if (!isPure(s.index) || !isPure(s.value)) return reject();
writes.add(s);
return everyStmtChild(s, expr, stmt);
}
const pushed = s.kind === "exprStmt" ? push(s.expr) : null;
if (pushed && !pushed.args.every(provenInt32)) {
if (!pushed.args.every(isPure)) return reject();
writes.add(s);
expr(pushed.receiver);
pushed.args.forEach(expr);
return !rejected;
}
return everyStmtChild(s, expr, stmt);
};
if (loop.cond) expr(loop.cond);
if (loop.kind === "for" && loop.update) stmt(loop.update);
for (const s of loop.body) stmt(s);
if (rejected || nodes > MAX_LOOP_NODES || arrays.size === 0) return null;
// A loop that rebinds one of the arrays could reach another array of
// either kind.
for (const id of arrays.keys()) if (assigned.has(id)) return null;
return { arrays, writes };
}
@@ -1,6 +1,6 @@
import type { IrExpr, IrType } from "../../ir/ir.js";
import type { IntegerRange } from "../../ir/integer-ranges.js";
import type { LlValue, LlvmEmitterContext } from "./expr-context.js";
import type { LlValue, LlvmEmitterContext, WideInteger } from "./expr-context.js";
export type LlInteger = NonNullable<LlValue["integer"]>;
@@ -51,6 +51,89 @@ export function exactInteger(
return { name, type: "i64", signed: true, range };
}
/** Doubles hold every integer of smaller magnitude exactly. */
const EXACT_DOUBLES = 2 ** 53;
/** Wide views stay within this magnitude: a product of two int32s reaches
* 2^62, one exact term of at most 2^53 may be added to it, and no i64
* result wraps. */
const WIDE_BOUND = 2 ** 62 + 2 ** 53;
/** Array and string indexes are below 2^32 - 1: a wide view exact below
* this many is exact at every index a length can admit. */
const INDEX_LIMIT = 2 ** 32;
function magnitude(range: IntegerRange): number {
return Math.max(Math.abs(range.min), Math.abs(range.max));
}
/** The wide view of `l op r` for `+`, `-` and `*` when the exact operands
* have no proven result range (their product or sum may pass 2^53).
*
* - A product of two exact integers equals its double whenever its
* magnitude is below 2^53: the double product of integers is exact there.
* - A sum or difference of two exact integers likewise.
* - A sum or difference of a wide view (exact below T) and an exact integer
* of magnitude at most B equals its double below min(T - B, 2^53): the
* view's magnitude is then below T, so its double is exact, and the
* double sum of two exact integers below 2^53 is exact.
*
* Every i64 result stays within 2^62 + 2^53 in magnitude, so none wraps. A view
* exact below fewer than 2^32 helps no index and is not kept. */
export function wideArithmetic(
host: LlvmEmitterContext,
op: string,
l: LlValue,
li: LlInteger | null,
r: LlValue,
ri: LlInteger | null,
): WideInteger | null {
if (op !== "+" && op !== "-" && op !== "*") return null;
const view = (value: LlValue, integer: LlInteger | null): WideInteger | null =>
integer
? { name: "", below: Infinity, bound: magnitude(integer.range) }
: (value.wideInteger ?? null);
const lv = view(l, li),
rv = view(r, ri);
if (!lv || !rv) return null;
let below: number, bound: number;
if (op === "*") {
if (lv.below !== Infinity || rv.below !== Infinity) return null;
below = EXACT_DOUBLES;
bound = lv.bound * rv.bound;
} else if (lv.below === Infinity && rv.below === Infinity) {
below = EXACT_DOUBLES;
bound = lv.bound + rv.bound;
} else if (lv.below === Infinity || rv.below === Infinity) {
const [exact, wide] = lv.below === Infinity ? [lv, rv] : [rv, lv];
below = Math.min(wide.below - exact.bound, EXACT_DOUBLES);
bound = exact.bound + wide.bound;
} else return null;
if (!(bound <= WIDE_BOUND) || below < INDEX_LIMIT) return null;
const operand = (value: LlValue, integer: LlInteger | null): string =>
integer ? widenInteger(host, integer) : value.wideInteger!.name;
const left = operand(l, li),
right = operand(r, ri);
const name = host.B.tmp();
host.B.line(
`${name} = ${op === "+" ? "add" : op === "-" ? "sub" : "mul"} nsw i64 ${left}, ${right}`,
);
return { name, below, bound };
}
/** An i64 index for a consumer that compares it, unsigned, with a length or
* capacity below 2^32 and takes the double's path otherwise: the exact
* integer of `value`, or its wide view, which equals the double at every
* such index. Null when the index needs the double round trip. */
export function indexInteger(
host: LlvmEmitterContext,
value: LlValue,
expr?: IrExpr,
): string | null {
const integer = exactInteger(host, value, expr);
if (integer) return widenInteger(host, integer);
const wide = value.wideInteger;
return wide && wide.below >= INDEX_LIMIT ? wide.name : null;
}
export function widenInteger(host: LlvmEmitterContext, value: LlInteger): string {
if (value.type === "i64") return value.name;
const wide = host.B.tmp();
@@ -973,6 +973,7 @@ export const LIB_FN_SYMS: Record<string, string> = {
"dyn.typedRefIs": "scr_dyn_typed_ref_is_key",
"dyn.classIs": "scr_dyn_class_is_key",
"dyn.iterPack": "scr_dyn_iter_pack",
"dyn.mathFold": "scr_dyn_math_fold",
"dyn.destructureRest": "scr_dyn_destructure_rest",
"dyn.iterator": "scr_dyn_iterator",
"dyn.asyncIteratorOpen": "scr_dyn_async_iterator_open",
@@ -1584,6 +1585,8 @@ export const LIB_FN_SYMS: Record<string, string> = {
"regex.resultProp": "scr_regex_result_prop",
"array.namedKeys": "scr_arr_named_keys",
"array.hasNamedKey": "scr_arr_has_named_key",
"array.hasProperties": "scr_arr_has_properties",
"array.undefinedKey": "scr_arr_undefined_key",
// queueMicrotask's checked-dynamic form (borrowed dyn; a non-function
// throws synchronously), the minted setImmediate value, and the
// timers/promises immediate — all mark the loop live.
@@ -13,7 +13,7 @@ import { analyzeCallLifetimes, type CallLifetimes } from "./call-lifetimes.js";
import type { LlValue, LlvmEmitterContext } from "./expr-context.js";
import { ReferenceEffects } from "./reference-effects.js";
import { SCR_ARR_FIELD, loadArrCount, scrArrStateAt } from "./common.js";
import { exactInteger, widenInteger } from "./integer-values.js";
import { indexInteger } from "./integer-values.js";
import { READ_FALLBACK } from "./dense-array-access.js";
import { compactRefLoadLines } from "./classes.js";
@@ -476,7 +476,8 @@ function emitDenseReferenceArrayRead(
no: string,
slow: string,
join: string,
loadType: "ptr" | "cref" | "i64" | "i8" = "ptr",
/** `i32`: an int32 array's slot, presented as the int32. */
loadType: "ptr" | "cref" | "i64" | "i8" | "i32" = "ptr",
): void {
const B = host.B;
const header = (field: number): string => {
@@ -567,7 +568,7 @@ function emitDenseReferenceArrayRead(
statePtr,
data,
cap,
loadType === "i8" ? 1 : loadType === "cref" ? 4 : 8,
loadType === "i8" ? 1 : loadType === "cref" || loadType === "i32" ? 4 : 8,
offset,
);
host.markMemoryPointer(statePtr, "array:present");
@@ -578,9 +579,10 @@ function emitDenseReferenceArrayRead(
B.startBlock(valueLabel);
const valuePtr = B.tmp(),
raw = B.tmp();
// Boolean arrays store one byte per element, compact references four,
// every other kind an 8-byte slot.
const slotType = loadType === "i8" ? "i8" : loadType === "cref" ? "i32" : "i64";
// Boolean arrays store one byte per element, compact references and int32
// arrays four, every other kind an 8-byte slot.
const slotType =
loadType === "i8" ? "i8" : loadType === "cref" || loadType === "i32" ? "i32" : "i64";
B.line(
`${valuePtr} = getelementptr inbounds ${slotType}, ptr ${data}, ${host.sizeType} ${offset}`,
);
@@ -638,11 +640,11 @@ export function emitBorrowedArrayRead(
return;
}
const index = host.emitExpr(read.index);
// A proven integer index (an int32 local, say) needs no double
// round-trip; a negative one compares above the capacity as unsigned.
const exact =
loopIndex === null && host.sizeType === "i64" ? exactInteger(host, index, read.index) : null;
const integerIndex = loopIndex ?? (exact ? widenInteger(host, exact) : null);
// A proven integer index (an int32 local, say), or a wide view exact at
// every index below the capacity, needs no double round-trip; a negative
// one compares above the capacity as unsigned.
const integerIndex =
loopIndex ?? (host.sizeType === "i64" ? indexInteger(host, index, read.index) : null);
const slow = B.newLabel("borrow.array.slow"),
join = B.newLabel("borrow.array.join");
emitDenseReferenceArrayRead(
@@ -680,13 +682,10 @@ export function emitLocalArrayRead(
const array = host.emitStableReceiver(read.array, [read.index]);
const loopIndex = inline ? host.emitIntegerLoopIndex(read.index) : null;
const index = host.emitExpr(read.index);
// A proven integer index (a counter with a known range, say) needs no
// double round trip, as in a borrowed read.
const exact =
inline && loopIndex === null && host.sizeType === "i64"
? exactInteger(host, index, read.index)
: null;
const integerIndex = loopIndex ?? (exact ? widenInteger(host, exact) : null);
// As in emitBorrowedArrayRead: an exact integer or wide view needs no
// round trip, and only the loop's own counter is an induction variable.
const integerIndex =
loopIndex ?? (inline && host.sizeType === "i64" ? indexInteger(host, index, read.index) : null);
const nullable = read.type.kind === "union" ? host.nullableUnions.get(read.type.unionId) : null;
if (nullable) {
// A nullable union is the element pointer itself: borrowed from an
@@ -743,8 +742,15 @@ export function emitLocalArrayRead(
const no = B.newLabel("local.array.missing"),
join = B.newLabel("local.array.join");
const scalar = isScalarElement(read.element);
// An int32 array's arms join on the int32 element, converted to the
// payload's double once at the join, so a consumer converting the payload
// back to an integer folds to the element.
const int32 = scalar && read.element.kind === "f64" && host.int32Array(read.array);
const element = int32 ? B.slot() : null;
if (element) B.entryAllocas.push(`${element} = alloca i32`);
const storeValue = (value: string): void => {
if (scalar) {
if (element) B.line(`store i32 ${value}, ptr ${element}`);
else if (scalar) {
// Slot bits for f64; a canonical 0/1 for bool, like scr_union_new_bool.
let bits = value;
if (read.element.kind === "bool") {
@@ -761,7 +767,37 @@ export function emitLocalArrayRead(
}
B.line(`store i32 ${read.presentTag}, ptr ${tag}`);
};
if (inline)
if (inline && scalar && host.checkedInt32Array(read.array))
// An int32 array with checked stores: its int32 arm stores the element's
// double bits, as the double arm does.
host.emitByElementKind(array.name, null, (kind) => {
const present =
kind === "i32"
? (value: string): void => {
const number = B.tmp(),
bits = B.tmp();
B.line(`${number} = sitofp i32 ${value} to double`);
B.line(`${bits} = bitcast double ${number} to i64`);
storeValue(bits);
}
: storeValue;
const loadType = kind === "i32" ? "i32" : "i64";
const induction = loopIndex !== null;
emitDenseReferenceArrayRead(
host,
array,
index,
integerIndex,
induction,
present,
no,
slow,
join,
loadType,
);
return null;
});
else if (inline)
emitDenseReferenceArrayRead(
host,
array,
@@ -772,7 +808,13 @@ export function emitLocalArrayRead(
no,
slow,
join,
!scalar ? referenceLoad(host, read.array.type) : read.element.kind === "bool" ? "i8" : "i64",
!scalar
? referenceLoad(host, read.array.type)
: read.element.kind === "bool"
? "i8"
: int32
? "i32"
: "i64",
);
else B.br(slow);
B.startBlock(slow);
@@ -799,7 +841,8 @@ export function emitLocalArrayRead(
B.line(
`${value} = call double @scr_arr_get_f64(ptr ${array.name}, double ${index.name})${READ_FALLBACK}`,
);
B.line(`${bits} = bitcast double ${value} to i64`);
if (element) B.line(`${bits} = fptosi double ${value} to i32`);
else B.line(`${bits} = bitcast double ${value} to i64`);
} else {
host.declare(`declare zeroext i1 @scr_arr_get_bool(ptr, double)`);
B.line(
@@ -807,7 +850,8 @@ export function emitLocalArrayRead(
);
B.line(`${bits} = zext i1 ${value} to i64`);
}
B.line(`store i64 ${bits}, ptr ${payload}`);
if (element) B.line(`store i32 ${bits}, ptr ${element}`);
else B.line(`store i64 ${bits}, ptr ${payload}`);
B.line(`store i32 ${read.presentTag}, ptr ${tag}`);
} else {
const value = B.tmp(),
@@ -821,10 +865,20 @@ export function emitLocalArrayRead(
}
B.br(join);
B.startBlock(no);
B.line(`store i64 0, ptr ${payload}`);
if (element) B.line(`store i32 0, ptr ${element}`);
else B.line(`store i64 0, ptr ${payload}`);
B.line(`store i32 ${read.missingTag}, ptr ${tag}`);
B.br(join);
B.startBlock(join);
if (element) {
const value = B.tmp(),
number = B.tmp(),
bits = B.tmp();
B.line(`${value} = load i32, ptr ${element}`);
B.line(`${number} = sitofp i32 ${value} to double`);
B.line(`${bits} = bitcast double ${number} to i64`);
B.line(`store i64 ${bits}, ptr ${payload}`);
}
B.line(`store ptr ${box}, ptr ${localSlot}`);
return read.borrow || scalar ? null : { slot: payload, type: read.element };
}
@@ -107,6 +107,7 @@ const INERT_LIB_CALLS = new Set([
"regex.resultProp",
"array.namedKeys",
"array.hasNamedKey",
"array.hasProperties",
"process.cwd",
"process.argv",
"process.execArgv",
@@ -35,12 +35,12 @@ export interface SpeculativeLoopPlan {
type SpeculativeLoop = Extract<IrStmt, { kind: "for" | "while" | "doWhile" }>;
/** The fast copy duplicates the loop; keep the duplicated code small. */
const MAX_LOOP_NODES = 400;
export const MAX_LOOP_NODES = 400;
/** Statements whose emission is plain control flow or a single write.
* Nested loops, switches and try statements keep the loop on the ordinary
* path (their emission keeps state that the two copies would not share). */
const SIMPLE_STATEMENTS = new Set<string>([
export const SIMPLE_STATEMENTS: ReadonlySet<string> = new Set<string>([
"varDecl",
"assign",
"exprStmt",
@@ -122,7 +122,8 @@ const COPIES = new Set([
const COPY_IN = new Set(["pushSpread", "unshiftSpread", "spliceInsert"]);
/** Library calls that only read array arguments. */
const READ_ONLY_LIB = /^(console\.|inspect\.|json\.stringify|math\.|insp\.)/;
const READ_ONLY_LIB =
/^(console\.|inspect\.|json\.stringify|math\.|insp\.|array\.(hasProperties|namedKeys|hasNamedKey)$)/;
/** Library calls that always throw (the value they stand for never exists). */
const THROWING_LIB = /^error\.\w*Throw$/;
@@ -872,6 +873,9 @@ export class Int32ElementContracts {
};
const isNumberArray = (t: IrType): boolean => t.kind === "array" && t.elem.kind === "f64";
// Live views handed straight to a read-only library call (util.inspect
// of an array with properties) are only read through.
const readOnlyViews = new Set<IrExpr>();
const expr = (e: IrExpr): void => {
switch (e.kind) {
case "assignExpr":
@@ -1000,6 +1004,7 @@ export class Int32ElementContracts {
case "libCall":
if (!READ_ONLY_LIB.test(e.fn))
for (const a of e.args) escape(a, a.loc, `the library call ${e.fn}`);
else for (const a of e.args) if (a.kind === "dynFrom") readOnlyViews.add(a);
break;
case "callValue":
for (const a of e.args) escape(a, a.loc, "a call through a function value");
@@ -1008,7 +1013,10 @@ export class Int32ElementContracts {
case "dynFrom":
// Plain arrays and records cross into dynamic code as deep copies;
// class instances and live references stay connected.
if (e.liveRef === true || this.containsObject(e.value.type, new Set()))
if (
!readOnlyViews.has(e) &&
(e.liveRef === true || this.containsObject(e.value.type, new Set()))
)
escape(e.value, e.loc, "dynamic (any or unknown) code");
break;
default:
@@ -34,6 +34,31 @@ export function packageShimRoot(): string {
: join(installedDeclarationRoot, "shims");
}
/** The directories of files scriptc itself ships, each with the prefix
* that names its files independently of where scriptc is installed
* (compilerOwnedFileName). IR artifacts spell source locations in them this
* way (serializeModuleParts). */
export function compilerOwnedFileRoots(): readonly { from: string; to: string }[] {
return [{ from: tsgoPath(packageShimRoot()) + "/", to: "scriptc:shims/" }];
}
/** The name of a file scriptc itself ships (a package shim or one of the
* shipped declaration files) that does not depend on where scriptc is
* installed: `scriptc:shims/<path below the shim root>` or
* `scriptc:<declaration file name>`. Null for every other file. Names and
* artifacts that would otherwise spell the installation's absolute path use
* it, so two installations of one compiler emit the same output. */
export function compilerOwnedFileName(file: string): string | null {
const path = tsgoPath(file);
for (const { from, to } of compilerOwnedFileRoots()) {
if (path.startsWith(from)) return to + path.slice(from.length);
}
for (const declaration of [ambientDtsPath(), overridesDtsPath(), fallbackDtsPath()]) {
if (path === declaration) return `scriptc:${path.slice(path.lastIndexOf("/") + 1)}`;
}
return null;
}
/** Path of the shipped ambient declarations — the always-shipped CORE
* (comptime, setTimeout). Part of EVERY program scriptc
* builds, the project-world preflight program included. */
@@ -60,6 +60,28 @@ test("semantic validation reads non-source dependencies once and still invalidat
expect(reads.mock.calls.filter(([path]) => path === dependency)).toHaveLength(1);
});
test("an equivalent edit of a file whose raw text the frontend read does not match", async () => {
const dir = await mkdtemp(join(tmpdir(), "scriptc-inputs-"));
scratch.push(dir);
const source = join(dir, "entry.ts");
const previous = "export const answer = 1;\n";
await writeFile(source, previous);
const tracker = new FrontendInputTracker();
tracker.run(() => {
trackedReadFile(source);
tracker.record({ op: "text", path: source });
});
const snapshot = tracker.snapshot();
expect(validFrontendInputSnapshot(snapshot)).toBe(true);
expect(frontendInputsStillMatch(snapshot)).toBe(true);
await writeFile(source, `// comment\n${previous}`);
const equivalent = vi.fn(() => true);
expect(
frontendInputsSemanticallyMatch(snapshot, new Map([[source, previous]]), equivalent),
).toBeNull();
expect(equivalent).not.toHaveBeenCalled();
});
test("tracked frontend reads invalidate on byte edits", async () => {
const dir = await mkdtemp(join(tmpdir(), "scriptc-inputs-"));
scratch.push(dir);
+51 -26
View File
@@ -2,6 +2,7 @@ import { createHash } from "node:crypto";
import { AsyncLocalStorage } from "node:async_hooks";
import { readFileSync, readdirSync, realpathSync, statSync } from "node:fs";
import { basename, dirname, resolve } from "node:path";
import { observeSourceText } from "./ts7/ast-node.js";
/**
* Filesystem observations made while constructing and lowering one frontend
@@ -16,7 +17,10 @@ export type FrontendInputProbe =
| { op: "kind"; path: string; kind: "file" | "directory" | "other" | "missing" }
| { op: "entries"; path: string; files: string[]; directories: string[] }
| { op: "entries-error"; path: string }
| { op: "realpath"; path: string; target: string | null };
| { op: "realpath"; path: string; target: string | null }
/** The frontend's output depends on the file's raw text or line
* positions, not only on its tokens (observeSourceText). */
| { op: "text"; path: string };
export interface FrontendInputSnapshot {
version: 1;
@@ -105,6 +109,7 @@ export class FrontendInputTracker {
record(probe: FrontendInputProbe): void {
const key = `${probe.op}\0${probe.path}`;
if (probe.op === "text" && this.probes.has(key)) return;
const previous = this.probes.get(key);
if (previous !== undefined && JSON.stringify(previous) !== JSON.stringify(probe)) {
// An input changed while this frontend was running. The build may
@@ -158,6 +163,13 @@ function record(probe: FrontendInputProbe): void {
(synchronousTracker ?? activeTracker.getStore())?.record(probe);
}
// Source text the frontend reads raw (call-site text in messages, comptime
// callbacks, line numbers) changes with edits that keep the tokens.
observeSourceText((fileName) => {
const tracker = synchronousTracker ?? activeTracker.getStore();
if (tracker !== undefined) tracker.record({ op: "text", path: resolve(fileName) });
});
/** Mark the active frontend as unsafe to persist because an exact host query
* bypassed the tracked filesystem wrappers. No-op outside a frontend run. */
export function markFrontendInputsUnstable(): void {
@@ -394,6 +406,9 @@ export function frontendInputsStillMatch(
return probe.target === null;
}
}
case "text":
// The file probe of the same path checks its content.
return true;
}
});
}
@@ -410,36 +425,44 @@ export function frontendInputsSemanticallyMatch(
exclusions: FrontendInputExclusions = {},
): FrontendSemanticMatch | null {
if (!validFrontendInputSnapshot(snapshot)) return null;
const sourceProbes = new Map(
snapshot.probes
.filter((probe): probe is Extract<FrontendInputProbe, { op: "file" }> => probe.op === "file")
.map((probe) => [probe.path, probe]),
);
const sourceProbes = new Map<string, Extract<FrontendInputProbe, { op: "file" }>>();
const rawText = new Set<string>();
for (const probe of snapshot.probes) {
if (probe.op === "file") sourceProbes.set(probe.path, probe);
else if (probe.op === "text") rawText.add(probe.path);
}
for (const [path, previous] of previousSources) {
if (sourceProbes.get(path)?.digest !== frontendSourceDigest(previous)) return null;
}
const changed: FrontendSemanticChange[] = [];
const currentSources = new Map<string, string>();
const adjusted: FrontendInputSnapshot = {
...snapshot,
probes: snapshot.probes.map((probe) => {
if (probe.op !== "file") return probe;
const previous = previousSources.get(probe.path);
if (previous === undefined || frontendSourceDigest(previous) !== probe.digest) return probe;
let current: string;
try {
current = readFileSync(probe.path, "utf8");
} catch {
return probe;
}
const currentDigest = frontendSourceDigest(current);
currentSources.set(probe.path, current);
if (currentDigest === probe.digest) return probe;
if (!isSemanticEquivalent(probe.path, previous, current)) return probe;
changed.push({ path: probe.path, previous, current });
return { ...probe, digest: currentDigest };
}),
};
const probes: FrontendInputProbe[] = [];
for (const probe of snapshot.probes) {
const previous = probe.op === "file" ? previousSources.get(probe.path) : undefined;
if (probe.op !== "file" || previous === undefined) {
probes.push(probe);
continue;
}
let current: string;
try {
current = readFileSync(probe.path, "utf8");
} catch {
return null;
}
const currentDigest = frontendSourceDigest(current);
currentSources.set(probe.path, current);
if (currentDigest === probe.digest) {
probes.push(probe);
continue;
}
// One edit that is not equivalent already rules out the match, and so
// does any edit of a file whose raw text the frontend read.
if (rawText.has(probe.path) || !isSemanticEquivalent(probe.path, previous, current))
return null;
changed.push({ path: probe.path, previous, current });
probes.push({ ...probe, digest: currentDigest });
}
const adjusted: FrontendInputSnapshot = { ...snapshot, probes };
if (!frontendInputsStillMatch(adjusted, exclusions)) return null;
for (const [path, previous] of previousSources) {
if (!currentSources.has(path)) {
@@ -486,6 +509,8 @@ export function validFrontendInputSnapshot(snapshot: unknown): snapshot is Front
return true;
case "realpath":
return value.target === null || typeof value.target === "string";
case "text":
return true;
default:
return false;
}
@@ -0,0 +1,123 @@
/* The string-keyed properties Node 24 exposes on its builtin constructor
* and namespace globals, own and inherited, so `"key" in Error` (a feature
* probe such as `"captureStackTrace" in Error`) answers statically. The
* global must be the unshadowed builtin; compiled programs cannot add
* properties to these objects, so the answer is fixed. */
const FUNCTION_PROTOTYPE = [
"length",
"name",
"arguments",
"caller",
"constructor",
"apply",
"bind",
"call",
"toString",
];
const OBJECT_PROTOTYPE = [
"constructor",
"__defineGetter__",
"__defineSetter__",
"hasOwnProperty",
"__lookupGetter__",
"__lookupSetter__",
"isPrototypeOf",
"propertyIsEnumerable",
"toString",
"valueOf",
"__proto__",
"toLocaleString",
];
const CONSTRUCTOR_OWN = ["length", "name", "prototype"];
const ERROR_OWN = [
...CONSTRUCTOR_OWN,
"captureStackTrace",
"prepareStackTrace",
"isError",
"stackTraceLimit",
];
/** Own string keys per global, beyond what the prototype chain adds. */
const CONSTRUCTORS: Readonly<Record<string, readonly string[]>> = {
Error: ERROR_OWN,
// The native error constructors inherit from Error.
TypeError: ERROR_OWN,
RangeError: ERROR_OWN,
SyntaxError: ERROR_OWN,
ReferenceError: ERROR_OWN,
EvalError: ERROR_OWN,
URIError: ERROR_OWN,
Object: [
...CONSTRUCTOR_OWN,
"assign",
"getOwnPropertyDescriptor",
"getOwnPropertyDescriptors",
"getOwnPropertyNames",
"getOwnPropertySymbols",
"hasOwn",
"is",
"preventExtensions",
"seal",
"create",
"defineProperties",
"defineProperty",
"freeze",
"getPrototypeOf",
"setPrototypeOf",
"isExtensible",
"isFrozen",
"isSealed",
"keys",
"entries",
"fromEntries",
"values",
"groupBy",
],
Array: [...CONSTRUCTOR_OWN, "isArray", "from", "fromAsync", "of"],
Promise: [
...CONSTRUCTOR_OWN,
"all",
"allSettled",
"any",
"race",
"resolve",
"reject",
"withResolvers",
"try",
],
Number: [
...CONSTRUCTOR_OWN,
"isFinite",
"isInteger",
"isNaN",
"isSafeInteger",
"parseFloat",
"parseInt",
"MAX_VALUE",
"MIN_VALUE",
"NaN",
"NEGATIVE_INFINITY",
"POSITIVE_INFINITY",
"MAX_SAFE_INTEGER",
"MIN_SAFE_INTEGER",
"EPSILON",
],
String: [...CONSTRUCTOR_OWN, "fromCharCode", "fromCodePoint", "raw"],
Map: [...CONSTRUCTOR_OWN, "groupBy"],
Set: CONSTRUCTOR_OWN,
WeakMap: CONSTRUCTOR_OWN,
WeakSet: CONSTRUCTOR_OWN,
Boolean: CONSTRUCTOR_OWN,
Function: CONSTRUCTOR_OWN,
Date: [...CONSTRUCTOR_OWN, "now", "parse", "UTC"],
BigInt: [...CONSTRUCTOR_OWN, "asUintN", "asIntN"],
};
/** Whether Node answers `key in <name>` with true, or null when `name` is
* not a tabled global. */
export function builtinGlobalHasKey(name: string, key: string): boolean | null {
const own = CONSTRUCTORS[name];
if (own === undefined) return null;
return own.includes(key) || FUNCTION_PROTOTYPE.includes(key) || OBJECT_PROTOTYPE.includes(key);
}
@@ -908,6 +908,11 @@ export class ClassDynamicDispatch {
private readonly prototypeClasses = new Set<string>();
private prototypeIdentityHelper: IrFunction | null = null;
private readonly prototypeIdentityClasses = new Set<string>();
/** `value instanceof Error` through Error.prototype: boxed instances of
* classes with a heritage evaluated at run time test their prototype
* data's chain. */
private errorPrototypeHelper: IrFunction | null = null;
private readonly errorPrototypeClasses = new Set<string>();
private readonly instancePrototypes = new Map<string, { fn: IrFunction; classes: Set<string> }>();
private readonly checkedCasts = new Map<string, { fn: IrFunction; classes: Set<string> }>();
/** The for-in key helpers, by libCall (dyn and native string[] keys). */
@@ -1114,6 +1119,49 @@ export class ClassDynamicDispatch {
});
changed = true;
}
if (
this.errorPrototypeHelper &&
!this.errorPrototypeClasses.has(className) &&
hasRuntimeHeritage(info)
) {
this.errorPrototypeClasses.add(className);
const loc = this.errorPrototypeHelper.loc;
const type: IrType = { kind: "object", className };
const value = varRef("value", DYN, loc);
const prototype = classPrototypeData(lowerer, info, loc, {
kind: "dynCheck",
value,
type,
loc,
});
if (prototype)
this.errorPrototypeHelper.body.unshift({
kind: "if",
cond: {
kind: "libCall",
fn: "dyn.typedRefIs",
args: [value, { kind: "strLit", value: typeKey(type), type: STRING, loc }],
type: BOOL,
loc,
},
then: [
{
kind: "return",
value: {
kind: "libCall",
fn: "dyn.inheritsErrorPrototype",
args: [prototype],
type: BOOL,
loc,
},
loc,
},
],
else_: null,
loc,
});
changed = true;
}
if (
this.serializesClasses &&
!this.jsonPrototypes.has(className) &&
@@ -1516,6 +1564,31 @@ export class ClassDynamicDispatch {
}
return { kind: "call", callee: cast.fn.name, args: [expr.value], type, loc: expr.loc };
}
if (expr.kind === "libCall" && expr.fn === "dyn.inheritsErrorPrototype") {
if (!this.errorPrototypeHelper) {
const loc = expr.loc;
this.errorPrototypeHelper = {
name: "%dyn.class.inheritsErrorPrototype",
params: [{ localId: "value", name: "value", type: DYN }],
locals: [{ id: "value", name: "value", type: DYN, mutable: false }],
returnType: BOOL,
loc,
body: [
{ kind: "return", value: { ...expr, args: [varRef("value", DYN, loc)] }, loc },
],
};
lowerer.liftedFns.push(this.errorPrototypeHelper);
this.generated.add(this.errorPrototypeHelper);
changed = true;
}
return {
kind: "call",
callee: this.errorPrototypeHelper.name,
args: expr.args,
type: BOOL,
loc: expr.loc,
};
}
if (
expr.kind === "libCall" &&
expr.fn === "dyn.getPrototype" &&
@@ -5942,3 +6015,11 @@ const INHERITED_OBJECT_METHODS: ReadonlySet<string> = new Set([
"hasOwnProperty",
"propertyIsEnumerable",
]);
/** Whether a class or one of its bases has a heritage evaluated at run
* time, whose prototype chain may reach Error.prototype. */
function hasRuntimeHeritage(info: ClassInfo): boolean {
for (let current: ClassInfo | null = info; current; current = current.base)
if (current.callableBase) return true;
return false;
}
@@ -380,6 +380,13 @@ export function collectClassMethodMutations(
markTarget(node.initializer, node);
if (ts.isPropertyAccessExpression(node) && isErrorPrototypeValue(lowerer, node))
lowerer.errorPrototypeValues = true;
// A heritage evaluated at run time may be Error, which links the
// class's instances to Error.prototype.
if (
(ts.isClassDeclaration(node) || ts.isClassExpression(node)) &&
runtimeHeritage(lowerer, node)
)
lowerer.errorPrototypeValues = true;
if (ts.isPropertyAccessExpression(node) || ts.isElementAccessExpression(node)) {
const base = unwrap(node.expression);
const prototype =
@@ -621,3 +628,22 @@ export function collectClassMethodMutations(
if (!receiverFields.has(field)) lowerer.receiverFreeCallbackFields.add(field);
}
}
/** Whether a class's `extends` expression is a value computed at run time:
* anything other than a name of a class declaration or of a builtin
* constructor (`extends Parent` over `const Parent = params.Parent ?? Object`,
* `extends make()`). */
function runtimeHeritage(
lowerer: Lowerer,
node: ts.ClassDeclaration | ts.ClassExpression,
): boolean {
const heritage = node.heritageClauses?.nodes.find(
(clause) => clause.token === ts.SyntaxKind.ExtendsKeyword,
)?.types?.nodes[0]?.expression;
if (heritage === undefined) return false;
if (!ts.isIdentifier(heritage) && !ts.isPropertyAccessExpression(heritage)) return true;
let symbol = lowerer.checker.getSymbolAtLocation(heritage);
if (symbol === undefined) return true;
if (symbol.flags & ts.SymbolFlags.Alias) symbol = lowerer.checker.getAliasedSymbol(symbol);
return !(symbol.flags & ts.SymbolFlags.Class) && !lowerer.isStdlibSymbol(symbol);
}
@@ -1,10 +1,10 @@
import * as ts from "../../ts7/adapter.js";
import { lowerOptionalIndex } from "../undefined-index.js";
import { locOf } from "../../program.js";
import { varRef } from "../../../ir/build.js";
import { DYN, type IrExpr, type IrStmt, isUnitType } from "../../../ir/ir.js";
import type { Lowerer } from "../lowerer.js";
import { completeFuncValueArgs } from "../call-arguments.js";
import { lowerOptionalNumber } from "../lower-exprs.js";
import { lowerSafeIndexRead } from "../array-values.js";
/** Retain native function slots during indexed calls. The array remains
@@ -37,7 +37,7 @@ export function lowerArrayCallbackCall(
return varRef(local.id, local.type, loc);
};
const receiver = store(array, "%callbackArray");
const index = lowerOptionalNumber(
const index = lowerOptionalIndex(
lowerer,
lowerer.lowerExpr(access.argumentExpression),
locOf(access.argumentExpression),
@@ -361,6 +361,21 @@ export function lowerArrayMethodCall(
if (probe.type.kind === "array") {
receiverIr = probe.type;
probedUntyped = true;
} else if (
probe.type.kind === "dyn" &&
!isJsSourceFile(ts.getSourceFileOfNode(access)) &&
lowerer.isStdlibMember(access)
) {
// An `unknown[]` (or `any[]`) array stored checked-dynamic — a
// generic body instantiated at `unknown`, a spread of unknown
// elements: the method runs on the runtime array (dynInvoke, JS
// exact), and a typed result is checked like any member read.
const dispatched = lowerDynDispatchMethodCall(lowerer, call, access, probe, true);
if (!dispatched) return null;
const result = lowerer.mapTypeOf(lowerer.typeOf(call));
return result && result.kind !== "void" && result.kind !== "dyn"
? lowerer.coerceInto(call, dispatched, result)
: dispatched;
}
}
if (receiverIr?.kind === "union") {
@@ -12,10 +12,11 @@ import { toPrimitiveHint } from "./class-dynamic-dispatch.js";
import { GENERATOR_STEP_METHODS, HELPER_METHODS } from "./generator-dynamic-dispatch.js";
/** Runtime entry points that step a checked-dynamic iterable's iterator:
* destructuring rest, and the collection constructors' and
* Object.fromEntries' seeds. */
* destructuring rest, the collection constructors' and Object.fromEntries'
* seeds, and Math.min/Math.max over a spread. */
const RUNTIME_STEP_LIB_FNS: ReadonlySet<string> = new Set([
"dyn.destructureRest",
"dyn.mathFold",
"dyn.nativeSetNew",
"dyn.mapSeedEntries",
"dyn.fromEntries",
@@ -80,6 +81,7 @@ const DISPATCH_LIB_CALLS = new Set<string>([
"dyn.propertyIsEnumerableComputed",
"dyn.defineProperty",
"dyn.getPrototype",
"dyn.inheritsErrorPrototype",
"dyn.reflectGet",
"dyn.reflectSet",
]);
@@ -464,8 +464,11 @@ function identityUse(root: ts.Node, name: string, from = -1): IdentityUse | null
cur = cur.expression;
return cur;
};
// The text appears only in warnings. A build that reported warnings is
// not reused after a source edit, so the read is not recorded as a source
// text dependency (ts.getTextOfNode would record it).
const textOf = (node: ts.Node): string => {
const text = ts.getTextOfNode(node);
const text = ts.getSourceFileOfNode(node).text.slice(ts.tokenStartOfNode(node), node.end);
return text.length > 60 ? `${text.slice(0, 57)}...` : text;
};
const visit = (node: ts.Node): "skip" | "stop" | undefined => {
@@ -1,6 +1,11 @@
/* Binary operators: arithmetic, comparisons and equality, logical and nullish pairs, and string conversion for concatenation. */
import { keyedHelperName } from "../stable-names.js";
import * as ts from "../../ts7/adapter.js";
import {
lowerOptionalIndex,
splitUndefinedIndex,
undefinedTypedArrayKeyFence,
} from "../undefined-index.js";
import { cleanFuncAdaptable } from "../coercions/function-adapters.js";
import { type Lowerer, jsFuncNameOf } from "../lowerer.js";
import {
@@ -1160,12 +1165,26 @@ export function lowerBinary(lowerer: Lowerer, expr: ts.BinaryExpression): IrExpr
}
if (recv?.type.kind === "array" || recv?.type.kind === "bytes") {
const receiver = lowerer.declareHiddenLocal("%setArray", recv.type);
const index = lowerOptionalNumber(
lowerer,
lowerer.lowerExpr(expr.left.argumentExpression),
locOf(expr.left.argumentExpression),
expr.left.argumentExpression,
);
// An undefined index names the "undefined" property
// (undefined-index.ts), which a typed array cannot hold.
const rawIndex = lowerer.lowerExpr(expr.left.argumentExpression);
const split =
recv.type.kind === "bytes"
? splitUndefinedIndex(
lowerer,
rawIndex,
expr.left.argumentExpression,
locOf(expr.left.argumentExpression),
)
: null;
const index =
split?.index ??
lowerOptionalIndex(
lowerer,
rawIndex,
locOf(expr.left.argumentExpression),
expr.left.argumentExpression,
);
if (index.type.kind !== "f64")
lowerer.unsupported(
"SC1090",
@@ -1178,7 +1197,7 @@ export function lowerBinary(lowerer: Lowerer, expr: ts.BinaryExpression): IrExpr
const result = varRef(value.id, raw.type, loc);
const arr = varRef(receiver.id, recv.type, loc);
const idx = varRef(key.id, F64, loc);
const write: IrStmt =
const store: IrStmt =
recv.type.kind === "array"
? arrayValueStore(lowerer, arr, idx, result, recv.type.elem, loc)
: {
@@ -1188,10 +1207,20 @@ export function lowerBinary(lowerer: Lowerer, expr: ts.BinaryExpression): IrExpr
value: lowerer.coerceInto(expr.right, result, F64),
loc,
};
const write: IrStmt = split
? {
kind: "if",
cond: split.isUndefined,
then: [undefinedTypedArrayKeyFence(lowerer, expr.left, loc)],
else_: [store],
loc,
}
: store;
return {
kind: "seqExpr",
stmts: [
{ kind: "varDecl", localId: receiver.id, init: recv, loc },
...(split?.stmts ?? []),
{ kind: "varDecl", localId: key.id, init: index, loc },
{ kind: "varDecl", localId: value.id, init: raw, loc },
write,
@@ -1,5 +1,10 @@
/* Element reads and writes (`a[i]`, `a[i] = v`), compound element updates, and keyed record reads. */
import * as ts from "../../ts7/adapter.js";
import {
lowerOptionalIndex,
splitUndefinedIndex,
undefinedTypedArrayKeyFence,
} from "../undefined-index.js";
import { type Lowerer, jsFuncNameOf, neverTaintedJsType } from "../lowerer.js";
import {
BOOL,
@@ -65,6 +70,7 @@ import { lowerClassPrototypeComputedAssignment } from "../class-prototypes.js";
import { isClassCallback } from "../class-callbacks.js";
import { isKeyedFieldAccess } from "./keyed-field-targets.js";
import { lowerFieldCompoundValue } from "./updates.js";
import { readsHeldEvolvingArray } from "../lower-stmts.js";
/** `a[i]` reads. Only f64 indices into array receivers are modeled; JS
* string-key element access (`a["length"]`) and string indexing (`s[0]`
@@ -751,12 +757,16 @@ export function lowerElementAccess(lowerer: Lowerer, expr: ts.ElementAccessExpre
: lowerer.maybeNarrow(arr, expr.expression);
}
// An alias or container of an evolving array whose reads disagree
// (`rows[0]` after `const rows = [xs]`) holds elements the checker's type
// does not list, so its element read stays checked-dynamic.
const heldEvolving = arr.type.kind === "dyn" && readsHeldEvolvingArray(lowerer, expr.expression);
// The checker sees an array but the VALUE lowered checked-dynamic (a
// jsdoc-typed default export from a .js module — signature 14): the
// read rides the dynCheck boundary when the array shape is one the checked-dynamic tree
// can validate, and fences by name when it is not — an arrayGet over a
// dyn receiver is never emitted (the validator ICE).
if (arr.type.kind === "dyn") {
if (arr.type.kind === "dyn" && !heldEvolving) {
if (
canDynCheckTo(
receiverIr,
@@ -774,14 +784,22 @@ export function lowerElementAccess(lowerer: Lowerer, expr: ts.ElementAccessExpre
}
}
// Array reads can themselves supply an index. A missing numeric value
// is undefined, whose property lookup misses just like NaN; keep that
// value instead of rejecting a checker-number index with optional IR.
const index = lowerOptionalNumber(
// is undefined, which names the array's "undefined" property
// (undefined-index.ts).
const index = lowerOptionalIndex(
lowerer,
lowerer.lowerExpr(expr.argumentExpression),
locOf(expr.argumentExpression),
expr.argumentExpression,
);
if (heldEvolving)
return {
kind: "dynKeyGet",
value: arr,
key: lowerer.coerceToExpected(index, DYN),
type: DYN,
loc: locOf(expr),
};
if (
arr.type.kind === "array" &&
(index.type.kind === "dyn" ||
@@ -1518,12 +1536,26 @@ export function lowerElementCompoundParts(
"compound assignment through a non-native array or byte view",
);
}
const index = lowerOptionalNumber(
lowerer,
lowerer.lowerExpr(target.argumentExpression),
locOf(target.argumentExpression),
target.argumentExpression,
);
// An undefined index names the "undefined" property (undefined-index.ts),
// which a typed array cannot hold.
const rawIndex = lowerer.lowerExpr(target.argumentExpression);
const split =
receiver.type.kind === "bytes"
? splitUndefinedIndex(
lowerer,
rawIndex,
target.argumentExpression,
locOf(target.argumentExpression),
)
: null;
const index =
split?.index ??
lowerOptionalIndex(
lowerer,
rawIndex,
locOf(target.argumentExpression),
target.argumentExpression,
);
if (index.type.kind !== "f64")
lowerer.unsupported("SC1090", target.argumentExpression, "indexing with non-number keys");
const receiverLocal = lowerer.declareHiddenLocal("%compoundArray", receiver.type);
@@ -1632,15 +1664,25 @@ export function lowerElementCompoundParts(
}
const resultLocal = lowerer.declareHiddenLocal("%compoundResult", computed.type);
const resultRef = (): IrExpr => varRef(resultLocal.id, computed.type, loc);
const write: IrStmt =
const store: IrStmt =
receiver.type.kind === "array"
? arrayValueStore(lowerer, receiverRef(), indexRef(), resultRef(), receiver.type.elem, loc)
: { kind: "bytesSet", arr: receiverRef(), index: indexRef(), value: resultRef(), loc };
const write: IrStmt = split
? {
kind: "if",
cond: split.isUndefined,
then: [undefinedTypedArrayKeyFence(lowerer, target, loc)],
else_: [store],
loc,
}
: store;
return {
expr: {
kind: "seqExpr",
stmts: [
{ kind: "varDecl", localId: receiverLocal.id, init: receiver, loc },
...(split?.stmts ?? []),
{ kind: "varDecl", localId: indexLocal.id, init: index, loc },
{ kind: "varDecl", localId: oldLocal.id, init: oldValue, loc },
{ kind: "varDecl", localId: resultLocal.id, init: computed, loc },
@@ -1867,13 +1909,23 @@ export function lowerElementWrite(lowerer: Lowerer, expr: ts.BinaryExpression):
// The checker's flow type is a typed array, but the storage is wider.
if (recv.type.kind !== "bytes") return lowerWiderBytesWrite(lowerer, expr, target, recv);
// An index the checker types as a number but whose storage may hold
// undefined (a hole read) converts with ToNumber, as the reads do.
const index = lowerOptionalNumber(
// undefined (a hole read) converts with ToNumber, as the reads do; an
// undefined index names a property the typed array cannot hold.
const rawIndex = lowerer.lowerExpr(target.argumentExpression);
const split = splitUndefinedIndex(
lowerer,
lowerer.lowerExpr(target.argumentExpression),
locOf(target.argumentExpression),
rawIndex,
target.argumentExpression,
locOf(target.argumentExpression),
);
const index =
split?.index ??
lowerOptionalNumber(
lowerer,
rawIndex,
locOf(target.argumentExpression),
target.argumentExpression,
);
if (index.type.kind === "dyn" && recv.type.kind === "bytes") {
return {
kind: "exprStmt",
@@ -1889,7 +1941,34 @@ export function lowerElementWrite(lowerer: Lowerer, expr: ts.BinaryExpression):
lowerer.unsupported("SC1090", target.argumentExpression, "indexing with non-number keys");
}
const value = lowerer.lowerExprExpecting(expr.right, F64);
return { kind: "bytesSet", arr: recv, index, value, loc: locOf(expr) };
if (split === null) return { kind: "bytesSet", arr: recv, index, value, loc: locOf(expr) };
const loc = locOf(expr);
const receiver = lowerer.declareHiddenLocal("%indexedBytes", recv.type);
const valueLocal = lowerer.declareHiddenLocal("%elementValue", F64);
return {
kind: "block",
body: [
{ kind: "varDecl", localId: receiver.id, init: recv, loc },
...split.stmts,
{ kind: "varDecl", localId: valueLocal.id, init: value, loc },
{
kind: "if",
cond: split.isUndefined,
then: [undefinedTypedArrayKeyFence(lowerer, target, loc)],
else_: [
{
kind: "bytesSet",
arr: varRef(receiver.id, recv.type, loc),
index,
value: varRef(valueLocal.id, F64, loc),
loc,
},
],
loc,
},
],
loc,
};
}
// Tuple element write `t[0] = v`: a positional-field write (recordSet)
// — literal indices only, like the read.
@@ -2184,7 +2263,7 @@ export function lowerElementWrite(lowerer: Lowerer, expr: ts.BinaryExpression):
"assignment through a possibly missing nested array receiver",
);
}
const index = lowerOptionalNumber(
const index = lowerOptionalIndex(
lowerer,
lowerer.lowerExpr(target.argumentExpression),
locOf(target.argumentExpression),
@@ -1,4 +1,5 @@
/* Identifier references in expression position: lowerExprInner's identifier branch and its helpers. */
import { predeclareForwardClass } from "../lower-stmts.js";
import { cjsMemberAliasValueOf } from "../cjs-exports-object.js";
import { builtinClassValue, builtinMemberClassKind } from "../builtins/builtin-class-dyn.js";
import * as ts from "../../ts7/adapter.js";
@@ -860,6 +861,12 @@ export function lowerIdentifierExpr(lowerer: Lowerer, expr: ts.Identifier, loc:
{
const sym = lowerer.resolveValueSymbol(expr);
const classInfo = sym ? lowerer.classBySymbol.get(sym) : undefined;
// A local class referenced from a function hoisted above its statement:
// its TDZ box, holding the class as a checked-dynamic constructor.
if (!classInfo && sym && predeclareForwardClass(lowerer, sym)) {
const local = lowerer.resolveLocal(expr);
if (local) return { kind: "varRef", localId: local.id, type: local.type, loc };
}
// A GENERIC class name whose CONTEXTUAL type pins one instantiation
// (`const B: new (v: string) => Counter<string> = Counter`): the
// value is that instantiation's class object — the generic-fn
@@ -37,6 +37,7 @@ import { fenceSymbolFieldCopy } from "../symbol-fields.js";
import { planUnionRetag } from "../union-retag.js";
import { jsHoldsDynObject, runtimeOptionalStorageOperand } from "../lower-exprs.js";
import { builtinGlobalValue } from "./property-access.js";
import { holdsEvolvingArray } from "../lower-stmts.js";
/** `{ a: 1, b: "x" }` → recordLit. The record type comes from the
* contextual type when tsc has one (annotated declarations, arguments,
@@ -397,7 +398,10 @@ export function lowerDynObjectLiteral(
? lowerer.lowerLambda(prop, true)
: ts.isShorthandPropertyAssignment(prop) && shorthandBuiltinValue(lowerer, prop)
? shorthandBuiltinValue(lowerer, prop)!
: !boxValue && ts.isObjectLiteralExpression(valueExpr)
: !boxValue &&
(ts.isObjectLiteralExpression(valueExpr) ||
(ts.isArrayLiteralExpression(valueExpr) &&
holdsEvolvingArray(lowerer, valueExpr)))
? lowerer.lowerExprExpecting(valueExpr, DYN)
: lowerer.lowerExpr(valueExpr as ts.Expression);
raw = fenceClosureProbe(lowerer, valueExpr, undefined, lowerValue) ?? lowerValue();
@@ -553,6 +553,19 @@ export function lowerPropertyAccessExpr(
expr.name.text,
)
) {
// A name the program stores on the global object
// (`(_a = globalThis).__config ?? (_a.__config = {})`) reads back from
// the global object's own properties.
const globalType = lowerer.typeOf(staticallySelected(lowerer, expr.expression));
if (globalThisWrittenNames(lowerer, globalType).has(expr.name.text)) {
return {
kind: "dynKeyGet",
value: lowerer.coerceToExpected(lowerer.lowerExpr(expr.expression), DYN),
key: { kind: "strLit", value: expr.name.text, type: STRING, loc },
type: DYN,
loc,
};
}
return dynUndefinedExpr(loc);
}
// `Uint8Array.prototype.toBase64` and its siblings, the base64/hex
@@ -4491,3 +4504,29 @@ function contextualArgumentCount(lowerer: Lowerer, fn: ts.Node): number | null {
}
return count;
}
/** The property names the program assigns on the global object, through
* `globalThis` or a binding of its type (`(_a = globalThis).x = v`). */
function globalThisWrittenNames(lowerer: Lowerer, globalType: ts.Type): Set<string> {
const known = lowerer.globalThisWrittenNames;
if (known) return known;
const names = new Set<string>();
lowerer.globalThisWrittenNames = names;
for (const sf of lowerer.program.getImplementationSourceFiles()) {
if (sf.isDeclarationFile || !sf.text.includes("globalThis")) continue;
ts.walkPreorder(sf, (node) => {
if (
ts.isBinaryExpression(node) &&
(node.operatorToken.kind === ts.SyntaxKind.EqualsToken ||
node.operatorToken.kind === ts.SyntaxKind.QuestionQuestionEqualsToken ||
node.operatorToken.kind === ts.SyntaxKind.BarBarEqualsToken ||
node.operatorToken.kind === ts.SyntaxKind.AmpersandAmpersandEqualsToken) &&
ts.isPropertyAccessExpression(node.left) &&
lowerer.typeOf(node.left.expression) === globalType
)
names.add(node.left.name.text);
return undefined;
});
}
return names;
}
@@ -51,6 +51,7 @@ import { isJsSourceFile, locOf } from "../../program.js";
import { checkedClassInstanceOf } from "../class-construction.js";
import { unsupportedDiag } from "../../../diagnostics/diagnostic.js";
import { globalSymbolKey } from "./global-symbols.js";
import { builtinGlobalHasKey } from "../builtins/global-namespace-keys.js";
import { foldedStringKeyOf } from "./object-literals.js";
import { NARROW_FIRST, sideEffectFreeOptionValue } from "../surfaces.js";
import { isExecExceptionType, refuseExecExceptionCode } from "../builtins/errors.js";
@@ -1430,6 +1431,16 @@ export function lowerInExpression(
"'in' with computed (non-literal) keys (a const whose type is one string literal folds; a string-typed key answers over index-signature record receivers)",
);
}
// A feature probe on a builtin constructor global (`"captureStackTrace"
// in Error`): the global is a fixed object, so the answer is static.
{
let recvNode = expr.right;
while (ts.isParenthesizedExpression(recvNode)) recvNode = recvNode.expression;
if (ts.isIdentifier(recvNode) && lowerer.isStdlibGlobal(recvNode, recvNode.text)) {
const has = builtinGlobalHasKey(recvNode.text, key);
if (has !== null) return { kind: "boolLit", value: has, type: BOOL, loc };
}
}
// A receiver re-read (the static folds below) is safe for plain reads
// AND for side-effect-free expressions (an object literal of literals
// — the `"a" in { a: true }` shape).
@@ -1,5 +1,10 @@
/* Increment, decrement and prefix unary operators, and compound field updates. */
import * as ts from "../../ts7/adapter.js";
import {
lowerOptionalIndex,
splitUndefinedIndex,
undefinedTypedArrayKeyFence,
} from "../undefined-index.js";
import { isStrictCode } from "../strict-mode.js";
import { isPrivateWriteReceiver } from "../private-brands.js";
import type { Lowerer } from "../lowerer.js";
@@ -311,12 +316,27 @@ export function lowerIncDec(
(raw.type.kind === "array" || raw.type.kind === "bytes")
) {
const receiver = save(raw, "%incrementArray");
const rawIndex = lowerOptionalNumber(
lowerer,
lowerer.lowerExpr(access.argumentExpression),
locOf(access.argumentExpression),
access.argumentExpression,
);
// An undefined index names the "undefined" property
// (undefined-index.ts), which a typed array cannot hold.
const loweredIndex = lowerer.lowerExpr(access.argumentExpression);
const split =
raw.type.kind === "bytes"
? splitUndefinedIndex(
lowerer,
loweredIndex,
access.argumentExpression,
locOf(access.argumentExpression),
)
: null;
if (split) stmts.push(...split.stmts);
const rawIndex =
split?.index ??
lowerOptionalIndex(
lowerer,
loweredIndex,
locOf(access.argumentExpression),
access.argumentExpression,
);
const index = save(
rawIndex.type.kind === "dyn" && isJsSourceFile(ts.getSourceFileOfNode(expr))
? { kind: "libCall", fn: "dyn.toNumberCoerce", args: [rawIndex], type: F64, loc }
@@ -340,7 +360,7 @@ export function lowerIncDec(
loc,
invalidNaN: true,
};
return finish(read, (value) =>
const store = (value: IrExpr): IrStmt =>
type.kind === "array"
? arrayValueStore(
lowerer,
@@ -356,8 +376,15 @@ export function lowerIncDec(
index,
value: lowerer.coerceInto(access, value, F64),
loc,
},
);
};
if (!split) return finish(read, store);
return finish(read, (value) => ({
kind: "if",
cond: split.isUndefined,
then: [undefinedTypedArrayKeyFence(lowerer, access, loc)],
else_: [store(value)],
loc,
}));
}
// A TS member statically typed number on a checked-dynamic receiver (a
// class-backed protocol view): the expanded form `o.v = o.v + 1`, its
@@ -15,6 +15,7 @@ import {
import { isJsSourceFile } from "../program.js";
import { PoisonError, dynFallbackType } from "./lowerer.js";
import { functionCanReturnUndefined } from "../function-completion.js";
import { jsBindingAppendsOnly, jsBindingHasOpenWrites } from "./lower-stmts.js";
import type { FnCtx } from "./function-context.js";
/** Keep the conservative completion rule shared by return construction
@@ -124,6 +125,32 @@ export function producesConstructor(lowerer: Lowerer, type: ts.Type): boolean {
return false;
}
/** Whether a return of `decl` itself (not of a nested function) is a bare
* read of a JavaScript variable whose writes the checker does not
* constrain. A string that is only appended to stays a string. */
function returnsOpenJsBinding(lowerer: Lowerer, decl: ts.SignatureDeclaration): boolean {
let open = false;
const opens = (value: ts.Expression | undefined): boolean => {
while (value && ts.isParenthesizedExpression(value)) value = value.expression;
if (!value || !ts.isIdentifier(value)) return false;
const binding = lowerer.resolveValueSymbol(value)?.valueDeclaration;
return (
binding !== undefined &&
ts.isVariableDeclaration(binding) &&
jsBindingHasOpenWrites(lowerer, binding) &&
!jsBindingAppendsOnly(lowerer, binding)
);
};
const scan = (node: ts.Node): void => {
if (open || (node !== decl && ts.isFunctionLike(node))) return;
if (ts.isReturnStatement(node) && opens(node.expression)) open = true;
ts.forEachChild(node, scan);
};
if (ts.isArrowFunction(decl) && decl.body && !ts.isBlock(decl.body)) open = opens(decl.body);
else scan(decl);
return open;
}
/** Map a declaration's return contract while retaining JavaScript
* inference fallbacks. Diagnostics preserve the caller's blame node. */
export function declaredReturnType(
@@ -212,6 +239,23 @@ export function declaredReturnType(
lowerer.unions.get(mappedReturn.unionId)?.arms.some((arm) => arm.kind === "func")
)
return DYN;
// A JS function returning a local that holds whatever its writes store
// (jsBindingHasOpenWrites): the inferred result is the initializer's
// type, which the returned value need not have.
// An async function's promise fulfills with that value.
if (
isJsSourceFile(ts.getSourceFileOfNode(decl)) &&
!decl.type &&
!decl.asteriskToken &&
!hasExplicitJsDocReturn(decl) &&
mappedReturn !== null &&
mappedReturn.kind !== "dyn" &&
!(mappedReturn.kind === "promise" && mappedReturn.inner.kind === "dyn") &&
returnsOpenJsBinding(lowerer, decl)
)
return decl.modifiers?.nodes.some((modifier) => modifier.kind === ts.SyntaxKind.AsyncKeyword)
? { kind: "promise", inner: DYN }
: DYN;
// JS inference assigns Number to arithmetic over any, although the
// runtime operands can both be BigInts. Preserve that native result.
if (
@@ -342,10 +342,15 @@ export function genericCallInstance(
const local = localClassResultInstance(lowerer, expr, info, rsig, params, retTs);
if (local) return local;
// A result instantiated at `any` (a type argument inferred from an
// untyped value, such as an untyped forwarding call) is checked-dynamic.
// untyped value, such as an untyped forwarding call) is checked-dynamic,
// and so is a JavaScript result whose JSDoc names a type the file never
// declares or imports (`@returns {MiddlewareHandler<E>}` in a published
// bundle): the checker's error type is documentation it could not
// resolve, not a contract.
const mappedReturn =
lowerer.mapTypeOf(retTs) ??
((retTs.flags & ts.TypeFlags.Any) !== 0 && !isJsSourceFile(ts.getSourceFileOfNode(info.decl))
((retTs.flags & ts.TypeFlags.Any) !== 0 &&
(!isJsSourceFile(ts.getSourceFileOfNode(info.decl)) || retTs.isErrorType())
? DYN
: null);
if (!mappedReturn) {
@@ -3797,18 +3797,22 @@ function lowerDynReceiverMethodCall(
}
if (access.name.text === "flatMap") return lowerDynArrayFlatMapCall(lowerer, call, access, recv);
// String methods claim only names NO other dyn-representable kind's
// prototype declares (Array carries includes/indexOf/slice too): for
// these, "the receiver is a string, or the call throws V8's TypeError"
// IS Node's semantics for every possible dyn value. Shared names would
// need a receiver-kind dispatch — they keep the fence.
// prototype declares (Array carries includes/indexOf/slice too), so no
// other kind inherits them. Shared names would need a receiver-kind
// dispatch — they keep the fence.
if (DYN_STRING_ONLY_METHODS.has(access.name.text) && !ownCallable) {
// An object-typed receiver the member is missing from (a JavaScript
// module object whose members are assigned in a function body,
// `common.repeat(" ", n)`) holds a string or an object with that
// member: dispatch on the receiver's runtime kind.
// A checked-dynamic object can still carry its OWN member of that
// name: zod's schemas assign `trim`, `toLowerCase` and `startsWith`
// in their constructors (`z.string().trim().min(1)`), an object's
// prototype can define `trim`, and a JavaScript module object can
// assign `repeat` in a function body (`common.repeat(" ", n)`, an
// object-typed receiver the member is missing from). An untyped
// receiver, or a member the checker cannot type, dispatches on the
// receiver's runtime kind: a string runs the string lowering, anything
// else calls its member as Node does.
if (
!checkerUntyped &&
(lowerer.typeOf(access).flags & (ts.TypeFlags.Any | ts.TypeFlags.Unknown)) !== 0 &&
(checkerUntyped ||
(lowerer.typeOf(access).flags & (ts.TypeFlags.Any | ts.TypeFlags.Unknown)) !== 0) &&
!call.questionDotToken &&
!access.questionDotToken &&
!call.arguments.nodes.some(ts.isSpreadElement)
@@ -4260,6 +4264,7 @@ const DYN_DISPATCH_METHODS = new Set([
"shift",
"unshift",
"slice",
"splice",
"at",
"indexOf",
"lastIndexOf",
@@ -8267,6 +8272,12 @@ function lowerObjectStaticCall(
...(fnName !== null ? { fnName } : {}),
};
}
// A class value (`Object.defineProperty(Definition, "name", { value })`)
// boxes to its constructor function, whose own-property table holds the
// definition for every alias of the class.
if (target?.type.kind === "classval" && lowerer.dynConvertible(target.type)) {
target = { kind: "dynFrom", value: target, type: DYN, loc: locOf(call.arguments.nodes[0]!) };
}
const descriptorPrimitive =
member === "getOwnPropertyDescriptor" &&
target &&
@@ -8662,11 +8673,15 @@ function lowerObjectStaticCall(
probed?.type.kind === "union" &&
isJsSourceFile(ts.getSourceFileOfNode(call)) &&
lowerer.dynConvertible(probed.type);
// An array walks its live view: the present indices, then its numeric
// and named properties.
const isArray = probed?.type.kind === "array" && lowerer.dynConvertible(probed.type);
if (
isDyn ||
isPrimitive ||
isFunction ||
isJsUnion ||
isArray ||
(probed && isDynTypedRefType(probed.type))
) {
const fn =
@@ -8676,7 +8691,14 @@ function lowerObjectStaticCall(
? "dyn.objValues"
: "dyn.objEntries";
let v = lowerer.lowerExpr(argNode);
if (v.type.kind !== "dyn") v = { kind: "dynFrom", value: v, type: DYN, loc: locOf(call) };
if (v.type.kind !== "dyn")
v = {
kind: "dynFrom",
value: v,
...(isArray ? { liveRef: true as const } : {}),
type: DYN,
loc: locOf(call),
};
// In TypeScript the keys come back as a native string[] and the
// entries as a native [string, any][]: the checker keeps every later
// write to the arrays well typed. JavaScript may store anything into
@@ -5,6 +5,7 @@ import {
isDatePrototypeMethod,
lowerDateSuperMethodCall,
} from "./builtins/date-classes.js";
import { predeclareForwardClass } from "./lower-stmts.js";
import { lowerDateStaticCall } from "./builtins/dates.js";
import { keyedHelperName } from "./stable-names.js";
import { lowerValueReturningConstructorNew } from "./js-dynamic-objects.js";
@@ -238,7 +239,7 @@ export function returnsOnlyThisValue(member: ts.MethodDeclaration): boolean {
return valid;
}
function functionLocalClass(decl: ts.ClassLikeDeclaration): boolean {
export function functionLocalClass(decl: ts.ClassLikeDeclaration): boolean {
for (
let parent: ts.Node | undefined = decl.parent;
parent && !ts.isSourceFile(parent);
@@ -2355,6 +2356,7 @@ export function collectClassShapeInner(
const checkedStatics =
isJsSourceFile(ts.getSourceFileOfNode(decl)) &&
((base && hasRuntimeStatics(base)) ||
definedThroughReflection(lowerer, decl) ||
decl.members.nodes.some(
(member) =>
ts.isAccessor(member) &&
@@ -5476,7 +5478,12 @@ export function classValueRef(lowerer: Lowerer, info: ClassInfo, blame: ts.Node)
const display = info.def.name.replace(/^%|^%m\d+\./, "");
fenceDecorationThrows(lowerer, info, blame);
if (info.localClass && ts.isIdentifier(blame)) {
const stored = lowerer.resolveLocal(blame);
let stored = lowerer.resolveLocal(blame);
// A reference from a function hoisted above the class statement.
if (!stored) {
const symbol = lowerer.resolveValueSymbol(blame);
if (symbol && predeclareForwardClass(lowerer, symbol)) stored = lowerer.resolveLocal(blame);
}
if (stored)
return lowerer.coerceToExpected(
{ kind: "varRef", localId: stored.id, type: stored.type, loc: locOf(blame) },
@@ -6638,7 +6645,7 @@ function staticCallOn(
// functions, called directly as `%C.static:m%n` — with the same
// through-a-VALUE shadowing fence as plain statics.
const gfound = findGenericStaticOn(lowerer, info, access.name.text);
if (!gfound) return null;
if (!gfound) return classConstructorCall(lowerer, call, access, info);
if (throughValue && staticShadowBelow(info, access.name.text)) {
lowerer.unsupported(
"SC1090",
@@ -6731,6 +6738,89 @@ function staticCallOn(
return { kind: "call", callee: fnName, args, type: found.method.ret, loc };
}
/** `C.call(thisArg, ...)` / `C.apply(thisArg, args)` on a program class that
* declares no such static: Function.prototype.call/apply invoke the class
* constructor without `new`, which throws Node's TypeError "Class
* constructor C cannot be invoked without 'new'" after the arguments are
* evaluated (and, for apply, after the argument list is read). The boxed
* class value's call path throws exactly that. Neither expando statics nor
* writes through untyped values can give a class an own `call` or `apply`.
* A receiver that may hold a subclass at run time (a class value, `this` in
* a static method) needs every subclass to lack such a static too. The
* receiver must be one the caller does not evaluate again (a name, `this`,
* or a namespace member). Null otherwise. */
function classConstructorCall(
lowerer: Lowerer,
call: ts.CallExpression,
access: ts.PropertyAccessExpression,
info: ClassInfo,
): IrExpr | null {
const name = access.name.text;
const receiver = access.expression;
if (
(name !== "call" && name !== "apply") ||
!info.decl ||
info.def.runtime ||
info.builtinError ||
info.builtinEmitter ||
info.builtinStream ||
info.builtinEventTarget ||
info.runtimeStatics?.length ||
declaresStaticAbove(info, name) ||
(exactClassOfReceiver(lowerer, receiver) !== info &&
info.subclasses.some((subclass) => declaresStaticBelow(subclass, name))) ||
call.arguments.nodes.some(ts.isSpreadElement) ||
!(
ts.isIdentifier(receiver) ||
receiver.kind === ts.SyntaxKind.ThisKeyword ||
(ts.isPropertyAccessExpression(receiver) && nsMemberIdentOf(lowerer, receiver) !== null)
)
)
return null;
return {
kind: "dynInvoke",
recv: lowerer.lowerExprExpecting(receiver, DYN),
method: name,
calleeName: ts.getTextOfNode(access),
args: call.arguments.nodes.map((argument) => lowerer.lowerExprExpecting(argument, DYN)),
type: DYN,
loc: locOf(call),
};
}
/** Whether `info`'s declaration has a static member that may be named `name`
* (a method, field or accessor of that name, or a computed name), or `info`
* has no declaration to tell. */
function declaresStatic(info: ClassInfo, name: string): boolean {
if (!info.decl) return true;
return info.decl.members.nodes.some(
(member) =>
member.name !== undefined &&
ts.canHaveModifiers(member) &&
ts.getModifiers(member)?.some((modifier) => modifier.kind === ts.SyntaxKind.StaticKeyword) ===
true &&
(ts.isComputedPropertyName(member.name) ||
((ts.isIdentifier(member.name) || ts.isStringLiteral(member.name)) &&
member.name.text === name)),
);
}
/** declaresStatic over `info` and its bases, which it inherits statics from. */
function declaresStaticAbove(info: ClassInfo, name: string): boolean {
for (let c: ClassInfo | null = info; c; c = c.base) if (declaresStatic(c, name)) return true;
return false;
}
/** declaresStatic over `info` and its subclasses, or runtime statics, which a
* function-local JavaScript class defines at run time. */
function declaresStaticBelow(info: ClassInfo, name: string): boolean {
return (
declaresStatic(info, name) ||
!!info.runtimeStatics?.length ||
info.subclasses.some((subclass) => declaresStaticBelow(subclass, name))
);
}
/** `C.m(...)` where `m` is a static method that a function-local
* JavaScript class defines at runtime (a factory's `class extends base`),
* inherited by `C` through its statically known base chain. When the body
@@ -11189,3 +11279,37 @@ function genericFieldFnNodeOf(
}
return null;
}
/** Whether a JavaScript class's own binding is the target of
* `Object.defineProperty`/`defineProperties` in its file
* (`Object.defineProperty(Definition, "name", { value })`): its constructor
* then needs a live own-property table. */
function definedThroughReflection(lowerer: Lowerer, decl: ts.ClassLikeDeclaration): boolean {
const name = decl.name;
const sf = ts.getSourceFileOfNode(decl);
if (name === undefined || !sf.text.includes("definePropert")) return false;
const symbol = lowerer.checker.getSymbolAtLocation(name);
if (symbol === undefined) return false;
let found = false;
ts.walkPreorder(sf, (node) => {
if (found) return "stop";
if (
ts.isCallExpression(node) &&
ts.isPropertyAccessExpression(node.expression) &&
(node.expression.name.text === "defineProperty" ||
node.expression.name.text === "defineProperties") &&
lowerer.stdlibGlobalMember(node.expression, "Object") !== null
) {
const target = node.arguments.nodes[0];
if (
target !== undefined &&
ts.isIdentifier(target) &&
target.text === name.text &&
lowerer.checker.getSymbolAtLocation(target) === symbol
)
found = true;
}
return undefined;
});
return found;
}
@@ -1,4 +1,10 @@
import { keyedHelperName } from "./stable-names.js";
import {
isNewTarget,
lowerNewTarget,
lowerNewTargetComparison,
lowerNewTargetName,
} from "./new-target.js";
import { lowerWebAssemblyValue, refuseWebAssemblyUse } from "./builtins/webassembly.js";
import { lowerAwaitedUniformPromiseAll } from "./builtins/promises.js";
import { dynUndefinedExpr, nodeThrowExpr, numLit, strLit, varRef } from "../../ir/build.js";
@@ -713,7 +719,8 @@ function lowerExprInner(lowerer: Lowerer, expr: ts.Expression): IrExpr {
if (ts.isPrefixUnaryExpression(expr)) return lowerer.lowerPrefixUnary(expr);
// `x++` / `x--` in expression position: yields the OLD value.
if (ts.isPostfixUnaryExpression(expr)) return lowerIncDec(lowerer, expr, false);
if (ts.isBinaryExpression(expr)) return lowerer.lowerBinary(expr);
if (ts.isBinaryExpression(expr))
return lowerNewTargetComparison(lowerer, expr) ?? lowerer.lowerBinary(expr);
if (ts.isCallExpression(expr)) {
// `h.return?.()` on a helper object: the method exists, so the
// optional call is the call.
@@ -729,13 +736,15 @@ function lowerExprInner(lowerer: Lowerer, expr: ts.Expression): IrExpr {
if (ts.isElementAccessExpression(expr)) return lowerer.lowerElementAccess(expr);
if (ts.isConditionalExpression(expr)) return lowerTernary(lowerer, expr);
if (ts.isPropertyAccessExpression(expr)) return lowerPropertyAccessExpr(lowerer, expr, loc);
if (ts.isPropertyAccessExpression(expr))
return lowerNewTargetName(lowerer, expr) ?? lowerPropertyAccessExpr(lowerer, expr, loc);
// Meta-properties, named: `new.target` reflects HOW a function was
// invoked (compiled functions are never constructors of themselves —
// no runtime invocation record exists), and `import.meta`/
// `import.defer` are module-loader surface a native binary does not
// carry.
if (isNewTarget(expr)) return lowerNewTarget(lowerer, expr);
if (ts.isMetaProperty(expr)) {
const name =
expr.keywordToken === ts.SyntaxKind.NewKeyword ? "new.target" : `import.${expr.name.text}`;
@@ -1,6 +1,5 @@
import * as ts from "../ts7/adapter.js";
import { DYN, STRING, arrayOf, type IrExpr } from "../../ir/ir.js";
import { strLit } from "../../ir/build.js";
import { locOf } from "../program.js";
import type { Lowerer } from "./lowerer.js";
import { TEMPORAL_GLOBAL } from "../builtin-modules.js";
@@ -19,10 +18,14 @@ export function lowerGlobalValue(lowerer: Lowerer, expr: ts.Expression): IrExpr
return {
kind: "libCall",
fn: "global.native",
// One immortal name table per program (keyed by its contents): reading
// the global object in a hot path allocates nothing, and the runtime
// recognizes a table it has already registered.
args: [
{
kind: "arrayLit",
elems: names.map((name) => strLit(name, loc)),
kind: "templateStrings",
key: `%globalThis.names:${names.join(",")}`,
cooked: names,
type: arrayOf(STRING),
loc,
},
@@ -1918,13 +1918,21 @@ function inspectHelper(lowerer: Lowerer, t: IrType, loc: SrcLoc): string {
// A closed record that kept the undeclared properties of its checked
// source renders through its live checked view, which holds them in the
// source's own-key order beside the current field values.
// source's own-key order beside the current field values. An array with
// properties besides its elements does too: the view lists them after
// the elements.
let prependSourceGate: (() => void) | null = null;
if (t.kind === "record" && recordMayKeepSource(lowerer, t)) {
const sourceTest: IrExpr | null =
t.kind === "record" && recordMayKeepSource(lowerer, t)
? recordSourceTest(v(), loc)
: t.kind === "array" && lowerer.dynConvertible(t)
? { kind: "libCall", fn: "array.hasProperties", args: [v()], type: BOOL, loc }
: null;
if (sourceTest !== null) {
prependSourceGate = (): void => {
body.unshift({
kind: "if",
cond: recordSourceTest(v(), loc),
cond: sourceTest,
then: [
ret({
kind: "libCall",
@@ -2496,6 +2496,25 @@ export function lowerStdlibMethodCall(
// method, drains into the number[] the fold reads.
src = classGeneratorIterable(lowerer, src) ?? src;
if (src.type.kind === "generator") src = generatorDrain(lowerer, src, F64, loc) ?? src;
// An untyped iterable (`Math.min(...caps.map(f))` in JavaScript): the
// runtime packs its values and folds their ToNumber like Node. V8 names
// a null or undefined source by its text, calls as `f(...)`.
if (src.type.kind === "dyn") {
const what = ts.isCallExpression(spread.expression)
? `${ts.getTextOfNode(spread.expression.expression)}(...)`
: ts.getTextOfNode(spread.expression);
return {
kind: "libCall",
fn: "dyn.mathFold",
args: [
src,
{ kind: "boolLit", value: name === "max", type: BOOL, loc },
{ kind: "strLit", value: what, type: STRING, loc },
],
type: F64,
loc,
};
}
if (src.type.kind !== "array" || src.type.elem.kind !== "f64") {
lowerer.unsupported(
"SC1090",
@@ -25,6 +25,7 @@ import { InternalCompilerError } from "../../errors.js";
* flow (if/while/for/for-of/do, switch, try/catch, abrupt completions),
* explicit resource management, and blocked-binding poisoning. */
import * as ts from "../ts7/adapter.js";
import { lowerOptionalIndex } from "./undefined-index.js";
import { processDataMemberValue, programReadsExitCode } from "./builtins/process.js";
import { requireInitFlagStmt } from "./require-initialization.js";
import { isJsonRequireDecl } from "./json-require.js";
@@ -204,7 +205,6 @@ import {
lowerElementCompoundStatement,
lowerIncDec,
lowerIncDecTarget,
lowerOptionalNumber,
lowerGroupsProjection,
matchResultNamedGroupsOf,
runtimeOptionalGuardIds,
@@ -1010,6 +1010,44 @@ export function predeclareForwardCapture(lowerer: Lowerer, symbol: ts.Symbol): b
return false;
}
/** The CLASS-DECLARATION twin of predeclareForwardCapture: a function
* hoisted above a local `class D` (`function make() { return new D(); }`
* reached through an earlier function) captures the class binding, which
* is in its temporal dead zone until the class statement runs. The class
* cannot be collected yet (its heritage may name bindings declared between
* here and the statement), so the binding is pre-declared as a TDZ box
* holding the class as a checked-dynamic constructor; the class statement
* later initializes it, and every reference reads it from there. */
export function predeclareForwardClass(lowerer: Lowerer, symbol: ts.Symbol): boolean {
const decl = symbol.valueDeclaration;
if (!decl || !ts.isClassDeclaration(decl) || decl.name === undefined) return false;
if (!isJsSourceFile(ts.getSourceFileOfNode(decl))) return false;
for (let i = lowerer.activeStmtLists.length - 1; i >= 0; i--) {
const entry = lowerer.activeStmtLists[i]!;
const idx = entry.stmts.indexOf(decl);
if (idx < 0) continue;
if (idx < entry.index || entry.ctx === lowerer.ctx || entry.ctx.tdzPredeclared.has(symbol))
return false;
const name = decl.name.text;
const count = entry.ctx.localCounters.get(name) ?? 0;
entry.ctx.localCounters.set(name, count + 1);
const local: IrLocal = {
id: `${name}.${count}`,
name,
type: DYN,
mutable: false,
boxed: true,
tdz: true,
};
entry.ctx.locals.push(local);
entry.frame.set(symbol, local);
entry.out.push({ kind: "varDecl", localId: local.id, init: null, loc: locOf(decl) });
entry.ctx.tdzPredeclared.set(symbol, local);
return true;
}
return false;
}
/** The FUNCTION-DECLARATION twin of predeclareForwardCapture: JS hoists a
* nested `function f() {}` to scope entry — the binding is LIVE from the
* first statement, so a reference lexically above the declaration in the
@@ -1220,6 +1258,43 @@ function lowerJsVarFromExisting(
};
}
/** Whether a JavaScript declaration's object value changes shape after its
* initializer (jsOpenObjectType, jsObjectBindingExpands) or carries
* methods, so only the checked-dynamic representation holds it.
* `inferred` is the declaration's mapped checker type. */
function jsDeclarationExpandsObject(
lowerer: Lowerer,
decl: ts.VariableDeclaration,
inferred: IrType | null,
): boolean {
const arms =
inferred?.kind === "union"
? (lowerer.unions.get(inferred.unionId)?.arms ?? [])
: inferred
? [inferred]
: [];
return (
(!decl.type &&
!hasJsTypeAnnotation(decl) &&
inferred !== null &&
jsOpenObjectType(decl, inferred, lowerer.shapes, lowerer.unions).kind === "dyn") ||
arms.some((arm) => arm.kind === "record" && jsObjectBindingExpands(lowerer, decl, arm)) ||
(isJsSourceFile(ts.getSourceFileOfNode(decl)) &&
!decl.type &&
!hasJsTypeAnnotation(decl) &&
decl.initializer !== undefined &&
ts.isObjectLiteralExpression(decl.initializer) &&
decl.initializer.properties.nodes.some(
(property) =>
ts.isMethodDeclaration(property) ||
ts.isAccessor(property) ||
(ts.isPropertyAssignment(property) &&
(ts.isFunctionExpression(property.initializer) ||
ts.isArrowFunction(property.initializer))),
))
);
}
/** The `var` symbol's function-scoped binding TYPE — the declared type at
* the binding name, with the JS-source fallbacks every mutable binding
* takes (an empty-object-literal type is checked-dynamic because tsc
@@ -1254,6 +1329,19 @@ function varBindingType(lowerer: Lowerer, nameNode: ts.Identifier): IrType | nul
type = staticImportNamespaceType(lowerer, nameNode.parent.initializer) ?? type;
}
if (!type || type.kind === "void") return null;
// A JavaScript `var` holds whatever its writes store, as a `let` does:
// the checker types it by its initializer and checks no later write
// (`var fraction = 0;`, later `fraction = match[7].slice(0, 3)`), and an
// object it holds may gain members (`mark.snippet = snippet(mark)`).
if (
type.kind !== "dyn" &&
ts.isVariableDeclaration(nameNode.parent) &&
nameNode.parent.name === nameNode &&
isJsSourceFile(ts.getSourceFileOfNode(nameNode)) &&
(jsBindingHasOpenWrites(lowerer, nameNode.parent) ||
jsDeclarationExpandsObject(lowerer, nameNode.parent, type))
)
return DYN;
// A binding readable before its first store holds undefined until then.
return lowerer.runtimeOptionalBindingType(nameNode, type);
}
@@ -1529,6 +1617,14 @@ export function lowerStmt(lowerer: Lowerer, stmt: ts.Statement): IrStmt | IrStmt
if (ts.isClassDeclaration(stmt)) {
if (!stmt.name) lowerer.unsupported("SC1090", stmt, "anonymous local class declarations");
const info = lowerer.lowerClassExpressionInfo(stmt);
// An earlier hoisted function captured the class: its TDZ box exists.
const symbol = lowerer.checker.getSymbolAtLocation(stmt.name);
const pre = symbol ? lowerer.ctx.tdzPredeclared.get(symbol) : undefined;
if (pre) {
lowerer.ctx.tdzPredeclared.delete(symbol!);
const value = lowerer.coerceToExpected(lowerer.lowerClassExpression(stmt), DYN);
return { kind: "assign", localId: pre.id, value, initializes: true, loc: locOf(stmt) };
}
const type: IrType = { kind: "classval", className: info.def.name };
const local = lowerer.declareLocal(stmt.name, stmt.name.text, type, true);
// Static members of a function-local class run while the class value is
@@ -4870,12 +4966,16 @@ export function bindPatternTarget(
}
// Preserve checked-value pattern bindings when the inferred type
// cannot represent them without a refusal or an alias-breaking copy.
// A name destructured from a container of an evolving array
// (evolvingArrayHold) may hold the array, which the checker types by
// the elements written before it.
const dynStays =
value.type.kind === "dyn" &&
(() => {
((() => {
const mapped = lowerer.mapTypeOf(lowerer.typeOf(name));
return mapped === null || jsBindingKeepsCheckedObject(lowerer, name, mapped);
})();
})() ||
readsHeldEvolvingArray(lowerer, name));
let type = dynStays
? DYN
: (lowerer.runtimeOptionalBindingType(name, lowerer.irTypeOf(name)) ??
@@ -5004,13 +5104,16 @@ function evolvingArrayType(lowerer: Lowerer, decl: ts.VariableDeclaration): IrTy
// type once the reads settle it.
const written: ts.Expression[] = [];
// The file's identifiers are indexed by name once (identifiersNamed), so
// each binding resolves only the uses spelled like it.
// each binding resolves only the uses spelled like it. A shorthand
// property (`{ xs }`) is a use too.
for (const node of identifiersNamed(lowerer, ts.getSourceFileOfNode(decl), decl.name.text)) {
if (
node === decl.name ||
node.pos < scope.pos ||
node.end > scope.end ||
lowerer.checker.getSymbolAtLocation(node) !== symbol
(ts.isShorthandPropertyAssignment(node.parent) && node.parent.name === node
? lowerer.checker.getShorthandAssignmentValueSymbol(node.parent)
: lowerer.checker.getSymbolAtLocation(node)) !== symbol
)
continue;
let target: ts.Node = node;
@@ -5068,28 +5171,105 @@ function evolvingArrayType(lowerer: Lowerer, decl: ts.VariableDeclaration): IrTy
return result;
}
/** An unannotated alias of an evolving array binding whose reads disagree
* (`const alias = xs` between `xs.push(1)` and `xs.push("two")`): the
* checker types the alias by the elements written before it, but both names
* hold one array, so the alias takes the binding's checked-dynamic type and
* sees the later writes. Aliases of such an alias follow it. Null for other
* bindings. */
function evolvingArrayAliasType(lowerer: Lowerer, decl: ts.VariableDeclaration): IrType | null {
let current = decl;
for (let depth = 0; depth < 8; depth++) {
let initializer = current.initializer;
while (initializer !== undefined && ts.isParenthesizedExpression(initializer))
initializer = initializer.expression;
if (initializer === undefined || !ts.isIdentifier(initializer)) return null;
const source = lowerer.checker.getSymbolAtLocation(initializer)?.valueDeclaration;
/** How a TypeScript expression holds an evolving array binding whose reads
* disagree (evolvingArrayType answers checked-dynamic): "evolving" for the
* binding itself, "alias" for an unannotated binding initialized with it or
* with an alias of it (`const alias = xs`), and "container" for an object or
* array literal with it as a property, element or spread source at any
* depth (`{ items: xs }`, `[xs]`), an unannotated binding initialized with
* or destructured from such a literal, and a property or element read of a
* container (`box.items`). The checker types an alias or container by the
* elements written before it was taken, but it holds the same array. Null
* for other expressions. */
function evolvingArrayHold(
lowerer: Lowerer,
expr: ts.Expression,
depth = 0,
): "evolving" | "alias" | "container" | null {
while (ts.isParenthesizedExpression(expr)) expr = expr.expression;
if (depth >= 8) return null;
const binding = (symbol: ts.Symbol | undefined): "evolving" | "alias" | "container" | null => {
const source = symbol?.valueDeclaration;
// A name destructured from a container (`const { items } = box`) may
// be the array or another container.
if (source !== undefined && ts.isBindingElement(source) && ts.isIdentifier(source.name)) {
const host = hostVariableDeclarationOf(source.name);
return host !== null &&
!host.type &&
host.initializer !== undefined &&
evolvingArrayHold(lowerer, host.initializer, depth + 1) === "container"
? "container"
: null;
}
if (source === undefined || !ts.isVariableDeclaration(source) || source.type) return null;
const evolved = evolvingArrayType(lowerer, source);
if (evolved !== null) return evolved.kind === "dyn" ? evolved : null;
current = source;
if (evolved !== null) return evolved.kind === "dyn" ? "evolving" : null;
if (source.initializer === undefined) return null;
const held = evolvingArrayHold(lowerer, source.initializer, depth + 1);
return held === "evolving" ? "alias" : held;
};
if (ts.isIdentifier(expr)) return binding(lowerer.checker.getSymbolAtLocation(expr));
if (ts.isPropertyAccessExpression(expr) || ts.isElementAccessExpression(expr))
return evolvingArrayHold(lowerer, expr.expression, depth + 1) === "container"
? "container"
: null;
// A spread of the array itself (`{ ...xs }`, `[...xs]`) copies its
// elements, as in Node; a spread of a container copies its references.
if (ts.isObjectLiteralExpression(expr)) {
for (const property of expr.properties.nodes) {
const held = ts.isPropertyAssignment(property)
? evolvingArrayHold(lowerer, property.initializer, depth + 1)
: ts.isShorthandPropertyAssignment(property)
? binding(lowerer.checker.getShorthandAssignmentValueSymbol(property))
: ts.isSpreadAssignment(property) &&
evolvingArrayHold(lowerer, property.expression, depth + 1) === "container"
? "container"
: null;
if (held !== null) return "container";
}
return null;
}
if (ts.isArrayLiteralExpression(expr)) {
for (const element of expr.elements.nodes) {
const held = ts.isSpreadElement(element)
? evolvingArrayHold(lowerer, element.expression, depth + 1) === "container"
? "container"
: null
: evolvingArrayHold(lowerer, element, depth + 1);
if (held !== null) return "container";
}
}
return null;
}
/** Whether a TypeScript expression is an alias or container of an evolving
* array binding whose reads disagree (evolvingArrayHold). Its value is
* checked-dynamic, and the checker's type for it lists only the elements
* written before it was taken, so element reads through it stay
* checked-dynamic instead of checking against that type. */
export function readsHeldEvolvingArray(lowerer: Lowerer, expr: ts.Expression): boolean {
if (isJsSourceFile(ts.getSourceFileOfNode(expr))) return false;
const held = evolvingArrayHold(lowerer, expr);
return held === "alias" || held === "container";
}
/** Whether a TypeScript object or array literal holds an evolving array
* binding whose reads disagree (evolvingArrayHold), so it builds as a
* checked-dynamic literal that keeps the array. */
export function holdsEvolvingArray(lowerer: Lowerer, expr: ts.Expression): boolean {
if (isJsSourceFile(ts.getSourceFileOfNode(expr))) return false;
return evolvingArrayHold(lowerer, expr) === "container";
}
/** An unannotated binding initialized with an evolving array binding whose
* reads disagree, or with an alias or container of one (evolvingArrayHold):
* it takes the array's checked-dynamic type, so it sees the later writes.
* Null for other bindings. */
function evolvingArrayHolderType(lowerer: Lowerer, decl: ts.VariableDeclaration): IrType | null {
if (decl.initializer === undefined) return null;
return evolvingArrayHold(lowerer, decl.initializer) === null ? null : DYN;
}
/** A single static type for an initializer-less JavaScript `let` whose
* later plain writes all agree. An evolving JS local reads as `undefined`
* until control flow proves a write, so a scalar slot is sound only when
@@ -5597,26 +5777,7 @@ export function lowerVarDecl(
: inferredObjectType
? [inferredObjectType]
: [];
const expandsObject =
(!decl.type &&
!hasJsTypeAnnotation(decl) &&
inferredObjectType !== null &&
jsOpenObjectType(decl, inferredObjectType, lowerer.shapes, lowerer.unions).kind === "dyn") ||
inferredObjectArms.some(
(arm) => arm.kind === "record" && jsObjectBindingExpands(lowerer, decl, arm),
) ||
(isJsSourceFile(ts.getSourceFileOfNode(decl)) &&
!decl.type &&
!hasJsTypeAnnotation(decl) &&
ts.isObjectLiteralExpression(decl.initializer) &&
decl.initializer.properties.nodes.some(
(property) =>
ts.isMethodDeclaration(property) ||
ts.isAccessor(property) ||
(ts.isPropertyAssignment(property) &&
(ts.isFunctionExpression(property.initializer) ||
ts.isArrowFunction(property.initializer))),
));
const expandsObject = jsDeclarationExpandsObject(lowerer, decl, inferredObjectType);
let init: IrExpr;
try {
init =
@@ -5803,7 +5964,7 @@ export function lowerVarDecl(
!decl.type &&
!isJsSourceFile(ts.getSourceFileOfNode(decl))
) {
const evolved = evolvingArrayType(lowerer, decl) ?? evolvingArrayAliasType(lowerer, decl);
const evolved = evolvingArrayType(lowerer, decl) ?? evolvingArrayHolderType(lowerer, decl);
if (evolved !== null) {
type = evolved;
init = lowerer.lowerExprExpecting(decl.initializer, evolved);
@@ -6327,18 +6488,15 @@ export function hasJsTypeAnnotation(decl: ts.VariableDeclaration): boolean {
);
}
/** Inferred JS scalar initializers do not constrain later writes through
* an any-valued callback. Keep those bindings open, including captured writes. */
export function jsBindingHasOpenWrites(lowerer: Lowerer, decl: ts.VariableDeclaration): boolean {
if (
!isJsSourceFile(ts.getSourceFileOfNode(decl)) ||
decl.type ||
hasJsTypeAnnotation(decl) ||
!ts.isIdentifier(decl.name)
)
return false;
const symbol = lowerer.resolveValueSymbol(decl.name);
if (!symbol) return false;
/** The assignments to a declaration's binding in its function (or file),
* `=` and compound alike. Null when the name has no symbol. */
function jsBindingWrites(
lowerer: Lowerer,
decl: ts.VariableDeclaration,
name: ts.Identifier,
): ts.BinaryExpression[] | null {
const symbol = lowerer.resolveValueSymbol(name);
if (!symbol) return null;
let owner: ts.Node = decl;
while (owner.parent && !ts.isFunctionLike(owner) && !ts.isSourceFile(owner)) owner = owner.parent;
let writes = lowerer.jsBindingWritesByOwner.get(owner);
@@ -6359,9 +6517,36 @@ export function jsBindingHasOpenWrites(lowerer: Lowerer, decl: ts.VariableDeclar
writes = indexed;
lowerer.jsBindingWritesByOwner.set(owner, writes);
}
const candidates = (writes.get(decl.name.text) ?? []).filter(
return (writes.get(name.text) ?? []).filter(
(node) => lowerer.resolveValueSymbol(node.left as ts.Identifier) === symbol,
);
}
/** Whether a JavaScript binding always holds a string: the checker types it
* as one, and every write appends to it (`result += part`), which yields a
* string whatever the part is. */
export function jsBindingAppendsOnly(lowerer: Lowerer, decl: ts.VariableDeclaration): boolean {
if (!ts.isIdentifier(decl.name)) return false;
if (lowerer.mapTypeOf(lowerer.typeOf(decl.name))?.kind !== "string") return false;
const writes = jsBindingWrites(lowerer, decl, decl.name);
return (
writes !== null &&
writes.every((node) => node.operatorToken.kind === ts.SyntaxKind.PlusEqualsToken)
);
}
/** Inferred JS scalar initializers do not constrain later writes through
* an any-valued callback. Keep those bindings open, including captured writes. */
export function jsBindingHasOpenWrites(lowerer: Lowerer, decl: ts.VariableDeclaration): boolean {
if (
!isJsSourceFile(ts.getSourceFileOfNode(decl)) ||
decl.type ||
hasJsTypeAnnotation(decl) ||
!ts.isIdentifier(decl.name)
)
return false;
const candidates = jsBindingWrites(lowerer, decl, decl.name);
if (candidates === null) return false;
if (
candidates.some(
(node) =>
@@ -6369,6 +6554,18 @@ export function jsBindingHasOpenWrites(lowerer: Lowerer, decl: ts.VariableDeclar
)
)
return true;
// A `+=` of a string turns a number or boolean binding into a string
// (`n += "px"`); the checker keeps the declared type.
if (
candidates.some(
(node) =>
node.operatorToken.kind === ts.SyntaxKind.PlusEqualsToken &&
(lowerer.typeOf(node.right).flags & ts.TypeFlags.StringLike) !== 0,
)
) {
const declared = lowerer.mapTypeOf(lowerer.typeOf(decl.name));
if (declared?.kind === "f64" || declared?.kind === "bool") return true;
}
// A write that can store null or undefined into a binding the checker
// types as a primitive (`offset = match[3] || null`): JavaScript files
// are checked without strict null checks, so the declared type does not
@@ -9212,7 +9409,7 @@ function lowerAssignTargetInto(
if (arr.type.kind === "array") {
const recv = lowerer.declareHiddenLocal("%dtRecv", arr.type);
out.push({ kind: "varDecl", localId: recv.id, init: arr, loc });
const index = lowerOptionalNumber(
const index = lowerOptionalIndex(
lowerer,
lowerer.lowerExpr(t.argumentExpression),
locOf(t.argumentExpression),
@@ -5683,7 +5683,10 @@ export function lowerStreamMethodCall(
// argument reaches an object-mode side's _write as passed, null when
// the callback took its place (a byte-mode side encodes a string
// chunk with it).
const writeEnc = (encArg: ts.Expression | undefined, cbArg: ts.Expression | undefined) => {
const writeEnc = (
encArg: ts.Expression | undefined,
cbArg: ts.Expression | undefined,
): IrExpr => {
const receiver = streamReceiverExpr(lowerer, access.expression);
const chunk = lowerer.lowerExprExpecting(args[0]!, DYN);
const tail: IrExpr[] = [];
@@ -5696,7 +5699,7 @@ export function lowerStreamMethodCall(
args: [receiver, numLit(flags, loc), chunk, ...tail],
type: BOOL,
loc,
} satisfies IrExpr;
};
};
if (kind === "dyn") {
if (args.length === 3) return writeEnc(args[1], args[2]);
@@ -1,5 +1,6 @@
import { isDeclBoundaryFile } from "../decl-projection.js";
import { relieveHeapPressure } from "../../heap-pressure.js";
import { moduleNamespaceObject } from "./module-namespace-elements.js";
import { keyedHelperName } from "./stable-names.js";
import { positionTag, StructuralPositions } from "./structural-positions.js";
import { onlyForwarded, stripTypeWrappers } from "./typed-arguments.js";
@@ -185,7 +186,7 @@ export { newFnCtx, type FnCtx } from "./function-context.js";
*/
import { dirname, relative, resolve } from "node:path";
import { bindingSource } from "../binding-source.js";
import { tsgoPath } from "../dts-paths.js";
import { compilerOwnedFileName, tsgoPath } from "../dts-paths.js";
import * as ts from "../ts7/adapter.js";
import { isBundledShimForwarder } from "../package-shims.js";
import { erasedTypeOnlyImport } from "../type-only-imports.js";
@@ -615,6 +616,7 @@ import {
lowerForStatement,
matchGroupsPatternSource,
isGuardedRecordRead,
holdsEvolvingArray,
} from "./lower-stmts.js";
import {
type FieldTarget,
@@ -2188,6 +2190,9 @@ export class Lowerer {
* and erased instances). */
erasingDepth = 0;
localClassInstantiations: { owner: ts.FunctionLikeDeclaration; name: string }[] = [];
/** The function-local classes of the program with an `extends` clause, whether lowering collected
* them yet or not (new-target.ts); found on first use. */
functionLocalSubclasses: ts.ClassLikeDeclaration[] | undefined = undefined;
/** Instance types (by checker type id) of the local classes generic
* factory calls return, and the per-instance class each names. */
readonly localClassInstanceTypes = new Map<number, string>();
@@ -2972,9 +2977,14 @@ export class Lowerer {
* ABI and shared prototype descriptors instead of call-site specialization. */
readonly prototypeMethodAccesses = new Map<string, ts.Node>();
/** Whether the program reads `Error.prototype`, the checked-dynamic
* object an ES5 constructor function can inherit from: only then can a
* value that is not an Error instance be `instanceof Error`. */
* object an ES5 constructor function can inherit from, or has a class
* whose heritage is evaluated at run time (`extends (Base ?? Error)`):
* only then can a value that is not an Error instance be
* `instanceof Error`. */
errorPrototypeValues = false;
/** The property names the program assigns on the global object
* (property-access.ts), collected on first use. */
globalThisWrittenNames: Set<string> | null = null;
/** Well-known symbol methods (`sym:iterator`) the program may replace
* through a symbol-keyed write: an instance or prototype property, or a
* reflection call. */
@@ -3201,13 +3211,14 @@ export class Lowerer {
/** Structural source positions (structural-positions.ts) for names that
* stay the same when unrelated code moves. */
/** Files are named by their path relative to the entry's directory, so
* names do not depend on where the program sits, and default library
* files by their base name: two checkers install them in different
* directories. */
* names do not depend on where the program sits, default library files
* by their base name (two checkers install them in different
* directories), and the files scriptc ships by their name inside scriptc,
* so names do not depend on where scriptc is installed either. */
readonly positions = new StructuralPositions((sf) =>
this.program.isSourceFileDefaultLibrary(sf)
? `lib:${sf.fileName.slice(sf.fileName.lastIndexOf("/") + 1)}`
: relative(dirname(this.entry.fileName), sf.fileName),
: (compilerOwnedFileName(sf.fileName) ?? relative(dirname(this.entry.fileName), sf.fileName)),
);
/** The outermost node whose token span is exactly `loc`, or null. */
@@ -3239,7 +3250,10 @@ export class Lowerer {
* before it. */
closureName(loc: SrcLoc, suffix = "", prefix = "%fn"): string {
const node = this.nodeAtLoc(loc);
const where = node === null ? `${loc.file}:${loc.start}:${loc.end}` : this.positions.of(node);
const where =
node === null
? `${compilerOwnedFileName(loc.file) ?? loc.file}:${loc.start}:${loc.end}`
: this.positions.of(node);
const key = `${prefix}|${this.closureRoot}|${where}|${suffix}`;
const first = keyedHelperName(this, prefix, key);
const seen = this.closureNameCounts.get(first) ?? 0;
@@ -3380,8 +3394,18 @@ export class Lowerer {
if (interned.has(fn.name)) this.sharedHelpers.add(fn);
if (this.sharedHelpers.size === 0) return;
const shared = new Map([...this.sharedHelpers].map((fn) => [fn.name, fn]));
const earlier = (left: SrcLoc, right: SrcLoc): boolean =>
left.file < right.file || (left.file === right.file && left.start < right.start);
// Files scriptc ships compare by their name inside scriptc, so the
// choice does not depend on where scriptc is installed.
const fileKeys = new Map<string, string>();
const fileKey = (file: string): string => {
let key = fileKeys.get(file);
if (key === undefined) fileKeys.set(file, (key = compilerOwnedFileName(file) ?? file));
return key;
};
const earlier = (left: SrcLoc, right: SrcLoc): boolean => {
if (left.file === right.file) return left.start < right.start;
return fileKey(left.file) < fileKey(right.file);
};
// The earliest use of each shared helper outside the shared helpers, and
// which shared helpers each shared helper uses.
const uses = new Map<string, SrcLoc>();
@@ -9465,6 +9489,22 @@ export class Lowerer {
}
if (expr.type.kind === "void" && expected.kind !== "void" && this.neverValued.has(expr))
return this.divergentValue(expr, expected);
// A builtin error constructor global stored in a slot of its class value
// type (`{ Parent: Error }`): the class value itself, which a checked
// cast of the dynamic constructor value could not produce.
if (
expected.kind === "classval" &&
expr.kind === "libCall" &&
expr.fn === "dyn.globalConstructor" &&
expr.args[0]?.kind === "strLit" &&
`%${expr.args[0].value}` === expected.className
) {
const info = this.classes.get(expected.className);
if (info?.builtinError && info.valueCtorParams !== undefined) {
this.defineClassValueConstruct(info);
return { kind: "classRef", className: info.def.name, type: expected, loc: expr.loc };
}
}
// process.stdout/stderr into a general Writable slot (general-streams.ts).
const stdioStream = stdioIntoGeneralStream(this, expr, expected);
if (stdioStream) return this.coerceToExpected(stdioStream, expected);
@@ -9571,6 +9611,13 @@ export class Lowerer {
// constructor. `require("events")` is the EventEmitter class itself,
// so its module binding is too (`ReleaseEmitter = EventEmitter` over
// `const EventEmitter = require("events")`).
if (expr.type.kind === "moduleNs" && isJsSourceFileName(expr.loc.file)) {
const object = moduleNamespaceObject(
this,
expr as IrExpr & { type: IrType & { kind: "moduleNs" } },
);
if (object) return object;
}
if (
expr.type.kind === "moduleNs" &&
expr.type.moduleId === "builtin:events" &&
@@ -10672,7 +10719,8 @@ export class Lowerer {
builtinOption("UrlObject") ||
builtinOption("URLFormatOptions") ||
!this.mapTypeOf(this.typeOf(x)) ||
this.excessJsContextualProperty(x, parts)
this.excessJsContextualProperty(x, parts) ||
holdsEvolvingArray(this, x)
) {
return lowerDynObjectLiteral(this, x);
}
@@ -1,6 +1,13 @@
import { dynUndefinedExpr, strLit, varRef } from "../../ir/build.js";
import * as ts from "../ts7/adapter.js";
import { DYN, STRING, type IrExpr, type IrFunction, type IrStmt } from "../../ir/ir.js";
import {
DYN,
STRING,
type IrExpr,
type IrFunction,
type IrStmt,
type IrType,
} from "../../ir/ir.js";
import { locOf } from "../program.js";
import { newFnCtx, type Lowerer } from "./lowerer.js";
@@ -25,12 +32,13 @@ export function lowerModuleNamespaceElement(
try {
const body: IrStmt[] = [];
const symbol = lowerer.checker.getSymbolAtLocation(source);
symbol?.getExports().forEach((exported, key) => {
for (const exported of symbol ? lowerer.checker.getExportsOfModule(symbol) : []) {
const key = ts.symbolName(exported);
const target =
exported.flags & ts.SymbolFlags.Alias
? lowerer.checker.getAliasedSymbol(exported)
: exported;
if (!(target.flags & ts.SymbolFlags.Value)) return;
if (!(target.flags & ts.SymbolFlags.Value)) continue;
const declaration = target.valueDeclaration;
const identifier = declaration?.name as ts.Node | undefined;
if (!identifier || !ts.isIdentifier(identifier)) {
@@ -43,7 +51,7 @@ export function lowerModuleNamespaceElement(
kind: "strEq",
negated: false,
left: varRef("key", STRING, loc),
right: strLit(String(key), loc),
right: strLit(key, loc),
type: { kind: "bool" },
loc,
},
@@ -51,7 +59,7 @@ export function lowerModuleNamespaceElement(
else_: null,
loc,
});
});
}
body.push({ kind: "return", value: dynUndefinedExpr(loc), loc });
const fn: IrFunction = {
name,
@@ -88,3 +96,78 @@ export function lowerModuleNamespaceElement(
};
return { kind: "call", callee: name, args: [receiver, text], type: DYN, loc };
}
/** A module namespace taken as an untyped value (`{ ...ns }`, `{ iso: ns }`
* in JavaScript): an object holding each value export, keyed in the
* namespace's sorted key order. The copy is exact while every export is a
* binding that never changes after initialization (const, function or
* class); null for other namespaces, which keep their fence. */
export function moduleNamespaceObject(
lowerer: Lowerer,
ns: IrExpr & { type: IrType & { kind: "moduleNs" } },
): IrExpr | null {
const source = lowerer.sourceFileOfModuleNamespace(ns.type);
if (!source) return null;
const symbol = lowerer.checker.getSymbolAtLocation(source);
if (!symbol) return null;
const exports: { key: string; identifier: ts.Identifier }[] = [];
let exact = true;
// The module's resolved exports, which include `export *` members.
for (const exported of lowerer.checker.getExportsOfModule(symbol)) {
const key = ts.symbolName(exported);
const target =
exported.flags & ts.SymbolFlags.Alias ? lowerer.checker.getAliasedSymbol(exported) : exported;
if (!(target.flags & ts.SymbolFlags.Value)) continue;
const declaration = target.valueDeclaration;
const identifier = declaration?.name as ts.Node | undefined;
const immutable =
declaration !== undefined &&
(ts.isFunctionDeclaration(declaration) ||
ts.isClassDeclaration(declaration) ||
(ts.isVariableDeclaration(declaration) &&
(ts.getCombinedNodeFlags(declaration) & ts.NodeFlags.Const) !== 0));
if (!identifier || !ts.isIdentifier(identifier) || !immutable) {
exact = false;
break;
}
exports.push({ key, identifier });
}
if (!exact) return null;
exports.sort((a, b) => (a.key < b.key ? -1 : a.key > b.key ? 1 : 0));
const name = `%module.object:${ns.type.moduleId}`;
const loc = ns.loc;
if (!lowerer.liftedFns.some((fn) => fn.name === name)) {
const context = newFnCtx(false, null, null, DYN);
const previousClass = lowerer.currentClass;
lowerer.currentClass = null;
lowerer.fnStack.push(context);
try {
const fields = exports.map(({ key, identifier }) => ({
key: strLit(key, loc),
value: lowerer.lowerExprExpecting(identifier, DYN),
}));
if (fields.some((field) => field.value.type.kind !== "dyn")) return null;
const fn: IrFunction = {
name,
params: [],
returnType: DYN,
locals: [...context.locals],
body: [{ kind: "return", value: { kind: "dynObjLit", fields, type: DYN, loc }, loc }],
loc,
};
lowerer.liftedFns.push(fn);
} finally {
lowerer.fnStack.pop();
lowerer.currentClass = previousClass;
}
}
const call: IrExpr = { kind: "call", callee: name, args: [], type: DYN, loc };
if (ns.kind === "moduleNsRef") return call;
return {
kind: "seqExpr",
stmts: [{ kind: "exprStmt", expr: ns, loc }],
result: call,
type: DYN,
loc,
};
}
@@ -0,0 +1,357 @@
/* `new.target`. Compiled code has no invocation record, but inside a class
* constructor `new.target` is always the class of the object under
* construction: a compiled `new` allocates the instance with its concrete
* class before any constructor of the chain runs, and every constructor of
* the chain receives that instance as `this`. The program's class
* hierarchy is closed, so the value is a static dispatch over the
* constructor's class and its descendants: the deepest class the instance
* is an instance of is its exact class.
*
* The checker rejects `new.target` in methods, accessors, static blocks and
* field initializers. Plain functions (constructible in JavaScript) keep
* the refusal. */
import * as ts from "../ts7/adapter.js";
import { BOOL, STRING, type IrExpr, type IrType, type SrcLoc } from "../../ir/ir.js";
import { varRef } from "../../ir/build.js";
import { locOf } from "../program.js";
import type { Lowerer } from "./lowerer.js";
import type { ClassInfo } from "./lower-classes.js";
import { classValueRef, functionLocalClass } from "./lower-classes.js";
export function isNewTarget(node: ts.Node): node is ts.MetaProperty {
return (
ts.isMetaProperty(node) &&
node.keywordToken === ts.SyntaxKind.NewKeyword &&
node.name.text === "target"
);
}
/** Where `new.target` is bound: a class constructor, or a plain function
* (refused). */
type Binding =
| { kind: "constructor"; declaration: ts.ConstructorDeclaration }
| { kind: "function" };
function bindingOf(node: ts.Node): Binding {
for (let p = node.parent; p !== undefined; p = p.parent) {
if (ts.isArrowFunction(p)) continue;
if (ts.isConstructorDeclaration(p)) return { kind: "constructor", declaration: p };
if (ts.isFunctionLike(p) || ts.isSourceFile(p)) return { kind: "function" };
}
return { kind: "function" };
}
interface Dispatch {
info: ClassInfo;
/** The constructor's descendants, deepest first, then its class. A
* generic family stands for its instantiations, also those that later
* bodies instantiate: its interval holds them all. */
classes: ClassInfo[];
self: IrExpr | null;
}
function depth(info: ClassInfo): number {
let d = 0;
for (let c = info.base; c; c = c.base) d++;
return d;
}
/** The function-local classes of the program that extend a class. */
function functionLocalSubclasses(lowerer: Lowerer): ts.ClassLikeDeclaration[] {
let found = lowerer.functionLocalSubclasses;
if (found === undefined) {
const classes: ts.ClassLikeDeclaration[] = [];
for (const sf of lowerer.moduleOrder)
ts.walkPreorder(sf, (node) => {
if (
(ts.isClassDeclaration(node) || ts.isClassExpression(node)) &&
baseExpressionOf(node) !== undefined &&
functionLocalClass(node)
)
classes.push(node);
});
lowerer.functionLocalSubclasses = found = classes;
}
return found;
}
function baseExpressionOf(node: ts.ClassLikeDeclaration): ts.Expression | undefined {
let base = node.heritageClauses?.nodes.find(
(clause) => clause.token === ts.SyntaxKind.ExtendsKeyword,
)?.types?.nodes[0]?.expression;
while (base !== undefined && ts.isParenthesizedExpression(base)) base = base.expression;
return base;
}
/** Whether the class `node` extends the class of `info` (for an
* instantiation, its family), directly or through other classes, as far as
* its `extends` clauses name classes. */
function derivesFrom(
lowerer: Lowerer,
node: ts.ClassLikeDeclaration,
info: ClassInfo,
seen: Set<ts.Node>,
): boolean {
if (seen.has(node)) return false;
seen.add(node);
const base = baseExpressionOf(node);
const name =
base === undefined
? undefined
: ts.isIdentifier(base)
? base
: ts.isPropertyAccessExpression(base)
? base.name
: undefined;
let symbol = name === undefined ? undefined : lowerer.checker.getSymbolAtLocation(name);
if (symbol !== undefined && (symbol.flags & ts.SymbolFlags.Alias) !== 0)
symbol = lowerer.checker.getAliasedSymbol(symbol);
if (symbol === undefined) return false;
const named = lowerer.classBySymbol.get(symbol);
if (named !== undefined)
return (
named === info ||
named === info.genericInstance?.family ||
lowerer.isSubclassOf(named.def.name, info.def.name)
);
// A class lowering has not collected: through its declaration.
for (const declaration of symbol.declarations) {
let cls: ts.Node | undefined = declaration;
if (ts.isVariableDeclaration(cls)) {
cls = cls.initializer;
while (cls !== undefined && ts.isParenthesizedExpression(cls)) cls = cls.expression;
}
if (
cls !== undefined &&
(ts.isClassDeclaration(cls) || ts.isClassExpression(cls)) &&
derivesFrom(lowerer, cls, info, seen)
)
return true;
}
return false;
}
/** `node`'s name, quoted, for diagnostics. */
function localClassName(node: ts.ClassLikeDeclaration): string {
if (node.name) return `'${node.name.text}'`;
const parent = node.parent;
return parent && ts.isVariableDeclaration(parent) && ts.isIdentifier(parent.name)
? `'${parent.name.text}'`
: "a class expression";
}
/** The classes `new.target` can denote in the constructor being lowered,
* or null after a diagnostic. */
function dispatchOf(lowerer: Lowerer, expr: ts.MetaProperty): Dispatch | null {
const binding = bindingOf(expr);
if (binding.kind !== "constructor") return null;
const info = lowerer.currentClass;
const cls = binding.declaration.parent;
if (!info || (info.decl !== null && info.decl !== cls)) {
lowerer.unsupported(
"SC1090",
expr,
"'new.target' in this constructor (its class has no lowering here)",
);
return null;
}
// An instantiation's subclasses can also be instantiations of generic
// subclasses of its family (`class G<T> extends Box<T>` under Box<number>).
const family = info.genericInstance?.family;
const descendants: ClassInfo[] = [];
for (const other of lowerer.classes.values()) {
if (other === info) continue;
if (
lowerer.isSubclassOf(other.def.name, info.def.name) ||
(family !== undefined &&
other.generic !== undefined &&
other !== family &&
lowerer.isSubclassOf(other.def.name, family.def.name))
)
descendants.push(other);
}
descendants.sort((a, b) => depth(b) - depth(a));
const classes = [...descendants, info];
for (const c of classes) {
if (c.localClass) {
lowerer.unsupported(
"SC1090",
expr,
`'new.target' where '${c.def.jsName ?? c.def.name}' is a function-local class (each evaluation creates a new class)`,
);
return null;
}
}
// Lowering collects a function-local class with the body declaring it,
// which may come after this constructor.
for (const local of functionLocalSubclasses(lowerer)) {
if (!derivesFrom(lowerer, local, info, new Set())) continue;
lowerer.unsupported(
"SC1090",
expr,
`'new.target' where ${localClassName(local)} is a function-local class (each evaluation creates a new class)`,
);
}
let self: IrExpr | null = null;
if (classes.length > 1) {
const thisLocal = lowerer.resolveThis(expr);
if (!thisLocal || thisLocal.type.kind !== "object") {
lowerer.unsupported(
"SC1090",
expr,
"'new.target' where the constructor's 'this' has no class representation",
);
return null;
}
self = varRef(thisLocal.id, thisLocal.type, locOf(expr));
}
return { info, classes, self };
}
/** A conditional chain selecting `arm(D)` for the exact class D of the
* instance under construction. */
function select(
dispatch: Dispatch,
type: IrType,
loc: SrcLoc,
arm: (info: ClassInfo) => IrExpr,
): IrExpr {
const { classes, self } = dispatch;
let result = arm(classes[classes.length - 1]!);
for (let i = classes.length - 2; i >= 0; i--) {
const c = classes[i]!;
result = {
kind: "ternary",
cond: { kind: "instanceOf", value: self!, className: c.def.name, type: BOOL, loc },
then: arm(c),
else_: result,
type,
loc,
};
}
return result;
}
function refuseFunction(lowerer: Lowerer, expr: ts.MetaProperty): never {
return lowerer.unsupported(
"SC1090",
expr,
"'new.target' in a plain function (a constructible function's invocation record does not exist in compiled code; use a class constructor)",
);
}
/** `new.target` as a value. */
export function lowerNewTarget(lowerer: Lowerer, expr: ts.MetaProperty): IrExpr {
const loc = locOf(expr);
if (bindingOf(expr).kind === "function") refuseFunction(lowerer, expr);
const dispatch = dispatchOf(lowerer, expr)!;
const type: IrType = { kind: "classval", className: dispatch.info.def.name };
for (const c of dispatch.classes) {
if (c.generic) {
lowerer.unsupported(
"SC1090",
expr,
`'new.target' as a value where '${c.def.jsName ?? c.def.name}' is a generic subclass (its instantiations have no common class value; read 'new.target.name' or compare 'new.target' with a class instead)`,
);
}
if (c !== dispatch.info && !lowerer.ctorAbiEquals(c, dispatch.info)) {
lowerer.unsupported(
"SC1090",
expr,
`'new.target' as a value where '${c.def.jsName ?? c.def.name}' constructs with a different signature than '${dispatch.info.def.jsName ?? dispatch.info.def.name}' (read 'new.target.name' or compare 'new.target' with a class instead)`,
);
}
}
return select(dispatch, type, loc, (c) =>
lowerer.coerceToExpected(classValueRef(lowerer, c, expr), type),
);
}
/** `new.target.name`: the exact class's name, without materializing a
* class value (subclasses may construct with other signatures). */
export function lowerNewTargetName(
lowerer: Lowerer,
access: ts.PropertyAccessExpression,
): IrExpr | null {
if (access.questionDotToken || access.name.text !== "name") return null;
let target: ts.Expression = access.expression;
while (ts.isParenthesizedExpression(target)) target = target.expression;
if (!isNewTarget(target) || bindingOf(target).kind !== "constructor") return null;
const dispatch = dispatchOf(lowerer, target);
if (!dispatch) return null;
const loc = locOf(access);
return select(dispatch, STRING, loc, (c) => {
// An own static `name` field replaces the class's name; a subclass has
// its own name. An instantiation's statics are its family's.
const owner = c.genericInstance?.family ?? c;
const field = owner.staticFields.find((f) => f.name === "name");
if (field !== undefined) {
if (field.type.kind !== "string")
lowerer.unsupported(
"SC1090",
access,
`'new.target.name' where '${c.def.jsName ?? c.def.name}' declares a static 'name' that is not a string`,
);
return { kind: "varRef", localId: field.globalId, type: field.type, loc };
}
if (owner.staticMethods?.has("name") || owner.genericStatics?.has("name"))
lowerer.unsupported(
"SC1090",
access,
`'new.target.name' where '${c.def.jsName ?? c.def.name}' declares a static method 'name'`,
);
// A generic family's instantiations share its name.
if (c.generic)
return { kind: "strLit", value: c.def.jsName ?? c.generic.baseName, type: STRING, loc };
return {
kind: "libCall",
fn: "class.name",
args: [classValueRef(lowerer, c, target)],
type: STRING,
loc,
};
});
}
/** `new.target === C` (and `!==`, `==`, `!=`, either operand order) with a
* program class named by the other operand: an exact-class test. */
export function lowerNewTargetComparison(
lowerer: Lowerer,
expr: ts.BinaryExpression,
): IrExpr | null {
const op = expr.operatorToken.kind;
const negated =
op === ts.SyntaxKind.ExclamationEqualsEqualsToken ||
op === ts.SyntaxKind.ExclamationEqualsToken;
if (
!negated &&
op !== ts.SyntaxKind.EqualsEqualsEqualsToken &&
op !== ts.SyntaxKind.EqualsEqualsToken
)
return null;
const strip = (e: ts.Expression): ts.Expression => {
while (ts.isParenthesizedExpression(e)) e = e.expression;
return e;
};
const left = strip(expr.left);
const right = strip(expr.right);
const target = isNewTarget(left) ? left : isNewTarget(right) ? right : null;
if (!target || bindingOf(target).kind !== "constructor") return null;
const other = target === left ? right : left;
if (!ts.isIdentifier(other)) return null;
// A constructor's new.target is never undefined.
if (other.text === "undefined" && (lowerer.typeOf(other).flags & ts.TypeFlags.Undefined) !== 0)
return { kind: "boolLit", value: negated, type: BOOL, loc: locOf(expr) };
const symbol = lowerer.resolveValueSymbol(other);
const named = symbol ? lowerer.classBySymbol.get(symbol) : undefined;
if (!named || named.generic || named.localClass) return null;
const dispatch = dispatchOf(lowerer, target);
if (!dispatch) return null;
const loc = locOf(expr);
return select(dispatch, BOOL, loc, (c) => ({
kind: "boolLit",
value: (c === named) !== negated,
type: BOOL,
loc,
}));
}
@@ -0,0 +1,101 @@
import * as ts from "../ts7/adapter.js";
import { BOOL, DYN, F64, UNDEFINED_T, type IrExpr, type IrStmt, type SrcLoc } from "../../ir/ir.js";
import { varRef } from "../../ir/build.js";
import type { Lowerer } from "./lowerer.js";
import { lowerOptionalNumber } from "./lower-exprs.js";
import { keyedHelperName } from "./stable-names.js";
/** An element index the checker types as a number can be undefined at run
* time (an element read from a holey array). JavaScript then names the
* property "undefined", not "NaN". */
/** The array element index `operand` (the lowered index of `node`) as a
* number. An undefined index is the runtime's "undefined" key
* (`array.undefinedKey`, a NaN the array's property table spells
* "undefined"); other values convert with ToNumber, as lowerOptionalNumber
* does. */
export function lowerOptionalIndex(
lowerer: Lowerer,
operand: IrExpr,
loc: SrcLoc,
node: ts.Expression,
): IrExpr {
const split = splitUndefinedIndex(lowerer, operand, node, loc);
if (split === null) return lowerOptionalNumber(lowerer, operand, loc, node);
return {
kind: "seqExpr",
stmts: split.stmts,
result: {
kind: "ternary",
cond: split.isUndefined,
then: { kind: "libCall", fn: "array.undefinedKey", args: [], type: F64, loc },
else_: split.index,
type: F64,
loc,
},
type: F64,
loc,
};
}
/** A typed-array element index that can be undefined: `stmts` evaluate it
* once, `isUndefined` tests it, and `index` is the number a defined index
* converts to. */
export interface SplitIndex {
stmts: IrStmt[];
isUndefined: IrExpr;
index: IrExpr;
}
/** Splits `raw`, the lowered index of `node`, when it is a number that can
* be undefined; null otherwise. */
export function splitUndefinedIndex(
lowerer: Lowerer,
raw: IrExpr,
node: ts.Expression,
loc: SrcLoc,
): SplitIndex | null {
if (raw.type.kind !== "union") return null;
const undefinedTag = lowerer.armTag(raw.type.unionId, UNDEFINED_T);
if (undefinedTag < 0) return null;
const local = lowerer.declareHiddenLocal("%elementIndex", raw.type);
const stable = varRef(local.id, raw.type, loc);
const index = lowerOptionalNumber(lowerer, stable, loc, node);
if (index.type.kind !== "f64") return null;
return {
stmts: [{ kind: "varDecl", localId: local.id, init: raw, loc }],
isUndefined: {
kind: "unionIsTag",
unionId: raw.type.unionId,
tag: undefinedTag,
negated: false,
value: stable,
type: BOOL,
loc,
},
index,
};
}
/** A typed array has no property named "undefined" to store into: the
* store throws the catchable SC1090 error when it runs. A call, so that it
* can sit among an expression's statements. */
export function undefinedTypedArrayKeyFence(lowerer: Lowerer, node: ts.Node, loc: SrcLoc): IrStmt {
const line = ts.getLineAndCharacterOfPosition(ts.getSourceFileOfNode(node), loc.start).line + 1;
const message = `typed array properties named 'undefined' (an element index that is undefined) are not supported yet [SC1090 at ${loc.file}:${line}]`;
const key = `index.undefinedFence:${message}`;
let name = lowerer.arrHofHelpers.get(key);
if (name === undefined) {
name = keyedHelperName(lowerer, "%index.undefinedFence", key);
lowerer.arrHofHelpers.set(key, name);
lowerer.liftedFns.push({
name,
params: [],
locals: [],
returnType: DYN,
body: [{ kind: "runtimeFence", code: "SC1090", message, loc }],
loc,
});
}
return { kind: "exprStmt", expr: { kind: "call", callee: name, args: [], type: DYN, loc }, loc };
}
@@ -0,0 +1,107 @@
import { expect, test } from "vitest";
import { tokenEquivalentSource } from "./token-equivalence.js";
const equivalent = (previous: string, current: string, path = "/src/a.ts") =>
tokenEquivalentSource(path, previous, current);
test("whitespace and plain comments are trivia", () => {
expect(equivalent("const a = 1;\n", "const a = 1; // one\n")).toBe(true);
expect(equivalent("f(a,b);\n", "f( a, /* first */ b );\n")).toBe(true);
expect(equivalent("const s = `x${a}y`;\n", "const s = `x${ a /* the value */ }y`;\n")).toBe(true);
});
test("tokens, literals and line breaks that change semicolon insertion are not", () => {
expect(equivalent("const a = 1;\n", "const a = 2;\n")).toBe(false);
expect(equivalent('const s = "a // b";\n', 'const s = "a // c";\n')).toBe(false);
expect(equivalent("const s = `a // ${b} /* c */`;\n", "const s = `a // ${b} /* d */`;\n")).toBe(
false,
);
// `return` followed by a line break returns undefined.
expect(equivalent("return a;\n", "return\na;\n")).toBe(false);
expect(equivalent("return a;\n", "return /*\n*/ a;\n")).toBe(false);
});
test("comment-like text inside regular expressions is code", () => {
expect(equivalent("const r = /[//]/; x = 1;\n", "const r = /[//]/; x = 2;\n")).toBe(false);
expect(equivalent("if (a) /\\/*/.test(b); c = 1;\n", "if (a) /\\/*/.test(b); c = 2;\n")).toBe(
false,
);
expect(equivalent("const q = a / b; // ratio\n", "const q = a / b; // the ratio\n")).toBe(true);
expect(equivalent("return /a/; // one\n", "return /a/; // two\n")).toBe(true);
// A `for await` header closes like a `for` header.
expect(
equivalent(
"for await (const x of xs) /[//]/.test(x); c = 1;\n",
"for await (const x of xs) /[//]/.test(x); c = 2;\n",
),
).toBe(false);
});
test("a / after a line break is never equivalent", () => {
const loop = (body: string) => `a: for (;;) {\n${body}\n}\n`;
// Automatic semicolon insertion decides whether these start a regular
// expression.
expect(
equivalent(loop("break\n/[//]/.test(x); c = 1;"), loop("break\n/[//]/.test(x); c = 2;")),
).toBe(false);
expect(
equivalent(loop("break a\n/[//]/.test(x); c = 1;"), loop("break a\n/[//]/.test(x); c = 2;")),
).toBe(false);
expect(
equivalent("type A = B\n/[//]/.test(x); c = 1;\n", "type A = B\n/[//]/.test(x); c = 2;\n"),
).toBe(false);
// A comment spanning lines separates them too; JavaScript keeps its comments as tokens.
expect(
equivalent(
"x\n/* a\n */ /[//]/.test(y); c = 1;\n",
"x\n/* a\n */ /[//]/.test(y); c = 2;\n",
"/src/a.js",
),
).toBe(false);
// A / on the line of the token before it still divides.
expect(equivalent(loop("continue\nx / 2; // one"), loop("continue\nx / 2; // two"))).toBe(true);
});
test("code units outside ASCII are equivalent only inside strings, templates and comments", () => {
expect(equivalent('const s = "é"; // one\n', 'const s = "é"; // two\n')).toBe(true);
expect(equivalent("const s = `é${a}`; // one\n", "const s = `é${a}`; // two\n")).toBe(true);
expect(equivalent("const a = 1; // é\n", "const a = 1; // è\n")).toBe(true);
expect(equivalent("const café = 1; // one\n", "const café = 1; // two\n")).toBe(false);
expect(
equivalent(
"ä: for (;;) { break ä\n/[//]/.test(x); c = 1; }\n",
"ä: for (;;) { break ä\n/[//]/.test(x); c = 2; }\n",
),
).toBe(false);
expect(equivalent("const r = /é/; // one\n", "const r = /é/; // two\n")).toBe(false);
});
test("whitespace outside ASCII is never equivalent", () => {
expect(
equivalent("return\u3000/[//]/.test(x); c = 1;\n", "return\u3000/[//]/.test(x); c = 2;\n"),
).toBe(false);
expect(equivalent("const a = 1;\u00a0// one\n", "const a = 1;\u00a0// two\n")).toBe(false);
});
test("U+2028 and U+2029 end a line comment, so their files are never equivalent", () => {
expect(equivalent("// a\u2028c = 1;\n", "// a\u2028c = 2;\n")).toBe(false);
expect(equivalent("/* a\u2029*/ c = 1;\n", "/* b\u2029*/ c = 1;\n")).toBe(false);
});
test("comments that can affect compilation are kept", () => {
expect(equivalent("/** @type {number} */ let a;\n", "/** @type {string} */ let a;\n")).toBe(
false,
);
expect(equivalent("// @ts-ignore\nf();\n", "f();\n")).toBe(false);
expect(equivalent('/// <reference path="a.d.ts" />\n', "\n")).toBe(false);
// Every comment is kept in JavaScript, whose types come from JSDoc.
expect(equivalent("let a; // one\n", "let a; // two\n", "/src/a.js")).toBe(false);
});
test("text the scanner cannot classify is never equivalent", () => {
expect(equivalent("x = {} / 2; // a\n", "x = {} / 2; // b\n")).toBe(false);
expect(equivalent("++/a/.lastIndex; // a\n", "++/a/.lastIndex; // b\n")).toBe(false);
expect(equivalent("const \\u0061 = 1; // a\n", "const \\u0061 = 1; // b\n")).toBe(false);
expect(equivalent("const a = <b/>; // a\n", "const a = <b/>; // b\n", "/src/a.tsx")).toBe(false);
expect(equivalent('const s = "open\n', 'const s = "open // \n')).toBe(false);
});
@@ -0,0 +1,391 @@
/* Token equivalence of two versions of a source file, for frontend cache
* reuse after edits that touch only comments and whitespace.
*
* Two versions are equivalent when they have the same tokens, in the same
* order, with line breaks between the same pairs of tokens (automatic
* semicolon insertion depends on them), and differ only in whitespace and
* in comments that cannot affect compilation. The scanner answers "not
* equivalent" whenever it cannot classify text with certainty, so a wrong
* answer can only cost a cache hit:
*
* - Every comment in a file that is not TypeScript (`.ts`, `.mts`, `.cts`)
* is kept as a token: JavaScript types come from JSDoc. TypeScript
* comments are kept when they start with `///` (references and
* directives) or contain `@` or `#` (checker directives, JSDoc tags, and
* annotations lowering reads).
* - A `/` starts a regular expression or divides according to the token
* before it. After `)` it divides unless the parenthesis closes an `if`,
* `while`, `for`, `for await`, or `with` header. After `}`, `++` or
* `--` the scanner cannot tell which, so it gives up, as it does for
* escaped identifiers, which can spell keywords.
* - A `/` with a line break (or a comment that spans lines) since the
* previous token gives up: whether automatic semicolon insertion ends the
* statement before it decides how it reads.
* - Template literals track their `${` substitutions on a brace stack.
* - Any code unit outside ASCII gives up, except inside strings, templates
* and comments. So do files that contain whitespace or a line terminator
* outside ASCII anywhere (U+2028 and U+2029 end a line comment), JSX
* files, and unterminated strings, comments, templates and regular
* expressions.
*
* The scanner models only what these rules name; anything else makes the
* versions not equivalent, which rebuilds. */
interface Token {
text: string;
comment: boolean;
lineBreakBefore: boolean;
}
/** Keywords after which `/` starts a regular expression. */
const EXPRESSION_KEYWORDS = new Set([
"return",
"typeof",
"instanceof",
"in",
"of",
"new",
"delete",
"void",
"throw",
"case",
"do",
"else",
"yield",
"await",
"default",
"extends",
]);
/** Whitespace and line terminators outside ASCII. */
const NON_ASCII_SPACE = /[\u00a0\u1680\u2000-\u200a\u2028\u2029\u202f\u205f\u3000\ufeff]/;
/** Keywords whose parenthesized header is followed by a statement. */
const HEADER_KEYWORDS = new Set(["if", "while", "for", "with"]);
const PUNCTUATORS = [
">>>=",
"...",
"===",
"!==",
"**=",
"<<=",
">>=",
">>>",
"&&=",
"||=",
"??=",
"=>",
"==",
"!=",
"<=",
">=",
"&&",
"||",
"??",
"?.",
"++",
"--",
"+=",
"-=",
"*=",
"%=",
"&=",
"|=",
"^=",
"**",
"<<",
">>",
];
function isTypeScript(path: string): boolean {
return path.endsWith(".ts") || path.endsWith(".mts") || path.endsWith(".cts");
}
function isIdentifierCode(code: number): boolean {
return (
(code >= 97 && code <= 122) ||
(code >= 65 && code <= 90) ||
(code >= 48 && code <= 57) ||
code === 36 ||
code === 95 ||
code === 92
);
}
function isLineBreak(code: number): boolean {
return code === 10 || code === 13;
}
function isSpace(code: number): boolean {
return code === 32 || code === 9 || code === 11 || code === 12 || isLineBreak(code);
}
/** The tokens of `source`, or null when the scanner cannot classify it. */
function scan(path: string, source: string): Token[] | null {
if (path.endsWith(".tsx") || path.endsWith(".jsx")) return null;
if (NON_ASCII_SPACE.test(source)) return null;
const typeScript = isTypeScript(path);
const tokens: Token[] = [];
// Brace kinds: true for a template substitution, false otherwise.
const braces: boolean[] = [];
// Paren kinds: true for a statement header.
const parens: boolean[] = [];
let lineBreakBefore = false;
// A line break, or a comment spanning lines, since the last token other
// than a comment.
let breakSinceCode = false;
// The previous token and the one before it, and whether a `)` closed a
// statement header.
const last: { token: string | null; before: string | null } = { token: null, before: null };
let closedHeader = false;
const push = (text: string, comment = false): void => {
tokens.push({ text, comment, lineBreakBefore });
lineBreakBefore = false;
if (!comment) {
breakSinceCode = false;
last.before = last.token;
last.token = text;
}
};
let index = 0;
const length = source.length;
// A shebang is one token.
if (source.startsWith("#!")) {
while (index < length && !isLineBreak(source.charCodeAt(index))) index++;
push(source.slice(0, index));
}
/** Scans template text from `index` (after a backtick or a closing
* substitution brace) through the closing backtick or `${`. */
const template = (start: number): boolean => {
let position = index;
while (position < length) {
const code = source.charCodeAt(position);
if (code === 92) {
position += 2;
continue;
}
if (code === 96) {
position++;
index = position;
push(source.slice(start, position));
closedHeader = false;
return true;
}
if (code === 36 && source.charCodeAt(position + 1) === 123) {
position += 2;
index = position;
push(source.slice(start, position));
braces.push(true);
last.before = last.token;
last.token = "${";
return true;
}
position++;
}
return false;
};
while (index < length) {
const code = source.charCodeAt(index);
if (isSpace(code)) {
if (isLineBreak(code)) lineBreakBefore = breakSinceCode = true;
index++;
continue;
}
// Outside strings, templates and comments, the scanner models ASCII only.
if (code > 127) return null;
const next = source.charCodeAt(index + 1);
// Comments.
if (code === 47 && (next === 47 || next === 42)) {
const start = index;
if (next === 47) {
while (index < length && !isLineBreak(source.charCodeAt(index))) index++;
} else {
const end = source.indexOf("*/", index + 2);
if (end < 0) return null;
index = end + 2;
}
const text = source.slice(start, index);
const semantic =
!typeScript || text.startsWith("///") || text.includes("@") || text.includes("#");
const multiline = /[\r\n]/.test(text);
if (multiline) breakSinceCode = true;
if (semantic) push(text, true);
else if (multiline) lineBreakBefore = true;
continue;
}
// Strings.
if (code === 34 || code === 39) {
const start = index;
index++;
while (index < length) {
const inner = source.charCodeAt(index);
if (inner === 92) {
index += 2;
continue;
}
if (isLineBreak(inner)) return null;
index++;
if (inner === code) break;
if (index >= length) return null;
}
if (source.charCodeAt(index - 1) !== code || index - start < 2) return null;
push(source.slice(start, index));
closedHeader = false;
continue;
}
// Templates.
if (code === 96) {
const start = index;
index++;
if (!template(start)) return null;
continue;
}
// Identifiers, keywords and numbers.
if (isIdentifierCode(code) || (code === 46 && next >= 48 && next <= 57)) {
const start = index;
index++;
while (index < length) {
const inner = source.charCodeAt(index);
if (isIdentifierCode(inner) || inner === 46) {
// A dot continues only a number (`1.5`), not member access.
if (inner === 46) {
const first = source.charCodeAt(start);
if (!((first >= 48 && first <= 57) || first === 46)) break;
}
index++;
continue;
}
// Exponent signs (`1e-5`).
if (
(inner === 43 || inner === 45) &&
(source.charCodeAt(index - 1) === 101 || source.charCodeAt(index - 1) === 69)
) {
const first = source.charCodeAt(start);
if ((first >= 48 && first <= 57) || first === 46) {
index++;
continue;
}
}
break;
}
const word = source.slice(start, index);
// An escaped identifier can spell a keyword (`\u0072eturn`).
if (word.includes("\\")) return null;
push(word);
closedHeader = false;
continue;
}
// Division or a regular expression.
if (code === 47) {
if (breakSinceCode) return null;
const before = last.token;
let regex: boolean;
if (before === null) regex = true;
else if (before === "}") return null;
else if (before === ")") regex = closedHeader;
// `++/re/.lastIndex` is valid, `x++ / 2` too.
else if (before === "++" || before === "--") return null;
else if (before === "]") regex = false;
else if (/^[A-Za-z0-9_$\\]/.test(before)) regex = EXPRESSION_KEYWORDS.has(before);
else if (before.startsWith("'") || before.startsWith('"') || before.endsWith("`"))
regex = false;
else if (before.startsWith("/") && before.length > 1 && before !== "/=") regex = false;
else regex = true;
if (regex) {
const start = index;
index++;
let inClass = false;
while (index < length) {
const inner = source.charCodeAt(index);
if (isLineBreak(inner) || inner > 127) return null;
if (inner === 92) {
index += 2;
continue;
}
if (inner === 91) inClass = true;
else if (inner === 93) inClass = false;
else if (inner === 47 && !inClass) break;
index++;
}
if (index >= length) return null;
index++;
while (index < length && isIdentifierCode(source.charCodeAt(index))) index++;
push(source.slice(start, index));
closedHeader = false;
continue;
}
const text = next === 61 ? "/=" : "/";
index += text.length;
push(text);
closedHeader = false;
continue;
}
// Brackets.
if (code === 40) {
parens.push(
last.token !== null &&
(HEADER_KEYWORDS.has(last.token) || (last.token === "await" && last.before === "for")),
);
index++;
push("(");
closedHeader = false;
continue;
}
if (code === 41) {
const header = parens.pop() ?? false;
index++;
push(")");
closedHeader = header;
continue;
}
if (code === 123) {
braces.push(false);
index++;
push("{");
closedHeader = false;
continue;
}
if (code === 125) {
if (braces.pop() === true) {
const start = index;
index++;
if (!template(start)) return null;
continue;
}
index++;
push("}");
closedHeader = false;
continue;
}
// Other punctuators.
let text = source[index]!;
for (const punctuator of PUNCTUATORS)
if (source.startsWith(punctuator, index)) {
text = punctuator;
break;
}
index += text.length;
push(text);
closedHeader = false;
}
if (braces.length !== 0 || parens.length !== 0) return null;
return tokens;
}
/** Whether `current` differs from `previous` only in whitespace and in
* comments that cannot affect compilation. */
export function tokenEquivalentSource(path: string, previous: string, current: string): boolean {
if (previous === current) return true;
const left = scan(path, previous);
if (left === null) return false;
const right = scan(path, current);
if (right === null || left.length !== right.length) return false;
for (let index = 0; index < left.length; index++) {
const a = left[index]!;
const b = right[index]!;
if (a.text !== b.text || a.comment !== b.comment) return false;
if (index > 0 && a.lineBreakBefore !== b.lineBreakBefore) return false;
}
return true;
}
+22 -1
View File
@@ -17,6 +17,16 @@ import {
import type { SourceFile } from "./ast-types.js";
import { SyntaxKind, NodeFlags } from "./enums.js";
/** Receives the file name of a source file whose raw text or line
* positions a reader takes: the text of a node other than a token
* (getText), a node's text with its trivia (getFullText), or line starts. */
let sourceTextObserver: ((fileName: string) => void) | undefined;
/** Installs the observer of raw source text reads (sourceTextObserver). */
export function observeSourceText(observer: ((fileName: string) => void) | undefined): void {
sourceTextObserver = observer;
}
/** A list of child nodes with the list's own position, the shape of tsc-ts's
* NodeList. The file keeps the nodes array, not this object: each read of a
* list field returns a new AstNodeList over the same array, so `nodes` keeps
@@ -210,6 +220,12 @@ export class AstFile {
return result;
}
/** Reports a read of this file's raw text or line positions. */
noteSourceText(): void {
if (sourceTextObserver !== undefined && this.root.kind === SyntaxKind.SourceFile)
sourceTextObserver(this.root.fileName);
}
/** The file name of the source file node at `index`; the root's is
* read once. */
sourceFileName(index: number): string {
@@ -220,6 +236,7 @@ export class AstFile {
}
lineStarts(): number[] {
this.noteSourceText();
if (this.lines === undefined) this.lines = astLineStarts(this.root.text ?? "");
return this.lines;
}
@@ -495,10 +512,14 @@ export class AstNode {
return this.getStart(sourceFile) - this.pos;
}
getFullText(sourceFile?: AstNode): string {
return ((sourceFile ?? this.file.root).text ?? "").substring(this.pos, this.end);
const source = sourceFile ?? this.file.root;
source.file.noteSourceText();
return (source.text ?? "").substring(this.pos, this.end);
}
getText(sourceFile?: AstNode): string {
const source = sourceFile ?? this.file.root;
// A token's text is the same in every token-equivalent version.
if (this.kind > SyntaxKind.LastToken) source.file.noteSourceText();
return (source.text ?? "").substring(this.getStart(source), this.end);
}
@@ -24,15 +24,11 @@ function writePackage(directory: string, name: string, executable?: string): voi
writeFileSync(join(directory, "lib", executable), "");
}
test("tsc-ts is the default checker except on Windows, and SCRIPTC_CHECKER selects one", () => {
expect(checkerBackend({}, "linux")).toBe("tsc-ts");
expect(checkerBackend({ SCRIPTC_CHECKER: "" }, "linux")).toBe("tsc-ts");
expect(checkerBackend({}, "darwin")).toBe("tsc-ts");
expect(checkerBackend({}, "win32")).toBe("tsgo");
expect(checkerBackend({ SCRIPTC_CHECKER: "" }, "win32")).toBe("tsgo");
expect(checkerBackend({ SCRIPTC_CHECKER: "tsc-ts" }, "linux")).toBe("tsc-ts");
expect(checkerBackend({ SCRIPTC_CHECKER: "tsc-ts" }, "win32")).toBe("tsc-ts");
expect(checkerBackend({ SCRIPTC_CHECKER: "tsgo" }, "linux")).toBe("tsgo");
test("tsc-ts is the default checker, and SCRIPTC_CHECKER selects one", () => {
expect(checkerBackend({})).toBe("tsc-ts");
expect(checkerBackend({ SCRIPTC_CHECKER: "" })).toBe("tsc-ts");
expect(checkerBackend({ SCRIPTC_CHECKER: "tsc-ts" })).toBe("tsc-ts");
expect(checkerBackend({ SCRIPTC_CHECKER: "tsgo" })).toBe("tsgo");
expect(() => checkerBackend({ SCRIPTC_CHECKER: "tsc" })).toThrow(
'SCRIPTC_CHECKER must be tsgo or tsc-ts, not "tsc"',
);
@@ -4,17 +4,13 @@ import { fileURLToPath } from "node:url";
/** The TypeScript 7 API servers a compile can check with: tsc-ts, the default,
* and tsgo from the `typescript` package, selected with `SCRIPTC_CHECKER=tsgo`.
* Both serve the same API and give the same results. On Windows the default is
* tsgo: the compiler connects to the server over a named pipe there, and tsc-ts's
* API server has no named-pipe transport yet. */
* Both serve the same API and give the same results. On Windows the Node host
* connects to either over a named pipe the server creates (rpc-process.ts). */
export type CheckerBackend = "tsgo" | "tsc-ts";
export function checkerBackend(
env: NodeJS.ProcessEnv = process.env,
platform: string = process.platform,
): CheckerBackend {
export function checkerBackend(env: NodeJS.ProcessEnv = process.env): CheckerBackend {
const value = env["SCRIPTC_CHECKER"];
if (value === undefined || value === "") return platform === "win32" ? "tsgo" : "tsc-ts";
if (value === undefined || value === "") return "tsc-ts";
if (value === "tsc-ts") return "tsc-ts";
if (value === "tsgo") return "tsgo";
throw new Error(`SCRIPTC_CHECKER must be tsgo or tsc-ts, not ${JSON.stringify(value)}`);
@@ -631,6 +631,14 @@ export class CheckerFacade {
return aliased;
}
/** A module symbol's exports, with the members of its `export *`
* declarations resolved (a symbol's own export table lists only the
* module's own declarations). */
getExportsOfModule(symbol: Ts7Symbol): readonly Ts7Symbol[] {
this.ensureActive();
return this.raw.getExportsOfModule(symbol);
}
getDeclaredTypeOfSymbol(symbol: Ts7Symbol): Type {
this.ensureActive();
let type = this.cache.declaredTypeOfSymbol.get(symbol);
+5 -4
View File
@@ -126,6 +126,7 @@ import { moduleRuntimeFeatures } from "./ir/runtime-features.js";
import { serializeModuleParts } from "./ir/serialize.js";
import { readTextParts, writeTextParts } from "./text-parts.js";
import { validateModule } from "./ir/validate.js";
import { compilerOwnedFileRoots } from "./frontend/dts-paths.js";
import { loadProgram } from "./frontend/program-node.js";
import { runFrontend, type FrontendFactory } from "./frontend/pipeline.js";
import {
@@ -786,7 +787,7 @@ async function prepareExecutableInput(
if (outputKind === "ir") {
await mkdir(dirname(opts.outPath), { recursive: true });
await writeTextParts(opts.outPath, serializeModuleParts(mod));
await writeTextParts(opts.outPath, serializeModuleParts(mod, compilerOwnedFileRoots()));
return { ok: true, artifact: { kind: "ir", path: opts.outPath } };
}
@@ -886,7 +887,7 @@ async function prepareExecutableInput(
if (opts.emitIr) {
irPath = defaultSourcePaths.ir;
irSourcePath = privateSiblingPath(irPath, "ir");
await writeTextParts(irSourcePath, serializeModuleParts(mod));
await writeTextParts(irSourcePath, serializeModuleParts(mod, compilerOwnedFileRoots()));
}
const nativeFeatures = executableNativeFeatures(mod, backend, opts.optimization ?? "release");
@@ -1987,7 +1988,7 @@ async function emitSemanticLibraryHit(
let irPath: string | undefined;
if (opts.emitIr) {
irPath = join(opts.outDir, `${stem}.lib.ir.json`);
await writeTextParts(irPath, serializeModuleParts(mod));
await writeTextParts(irPath, serializeModuleParts(mod, compilerOwnedFileRoots()));
}
await compileLibraryNative(profile, llvmPath, archivePath, opts.sanitize ?? false, hit.native);
timing("native-archive");
@@ -2273,7 +2274,7 @@ async function compileLibraryTracked(
let irPath: string | undefined;
if (opts.emitIr) {
irPath = join(opts.outDir, `${stem}.lib.ir.json`);
await writeTextParts(irPath, serializeModuleParts(mod));
await writeTextParts(irPath, serializeModuleParts(mod, compilerOwnedFileRoots()));
}
const nativeFeatures = libraryNativeFeatures(mod, profile.emission);
@@ -672,6 +672,8 @@ export const MAY_THROW_LIB_FNS: ReadonlySet<IrLibFn> = new Set([
"dyn.propertyIsEnumerableComputed",
// the destructuring pack throws V8's TypeError on non-iterable dyn kinds
"dyn.iterPack",
// ToNumber of each element can run valueOf
"dyn.mathFold",
"dyn.destructureRest",
"dyn.arrayFromIterator",
"dyn.iterator",
@@ -219,6 +219,7 @@ export const LIB_FN_SIGS = defineLibFnSignatures({
* not iterable (cannot read property Symbol(Symbol.iterator))"). In the
* may-throw seed set. */
"dyn.iterPack": { argTypes: [DYN, STRING], result: DYN },
"dyn.mathFold": { argTypes: [DYN, BOOL, STRING], result: F64 },
"dyn.destructureRest": { argTypes: [DYN, STRING], result: DYN },
"dyn.arrayFromIterator": { argTypes: [DYN], result: DYN },
"dyn.iterator": { argTypes: [DYN, STRING], result: DYN },
@@ -3821,6 +3822,13 @@ export const LIB_FN_SIGS = defineLibFnSignatures({
// Whether the array (args[0]) still has the key (args[1]) from its
// namedKeys list. Borrowed; never throws.
"array.hasNamedKey": { argTypes: [null, STRING], result: BOOL },
// Whether an array (args[0], any element type) has properties besides its
// elements: noncanonical numeric keys, match metadata or named
// properties. Borrowed; never throws.
"array.hasProperties": { argTypes: [null], result: BOOL },
// The key of an element access whose index is undefined (the property
// named "undefined"; frontend/lowering/undefined-index.ts).
"array.undefinedKey": { argTypes: [], result: F64 },
// node:events EventEmitter: receivers are emitter-hierarchy objects and
// the chaining forms (on/off/removeAll/setMax) return the receiver's
// own class — program-dependent object types, checked in the libCall
@@ -161,6 +161,10 @@ function unitLocal(local: IrLocal): IrLocal {
* declaration from a one-argument charAt, a one-unit test, or a switch
* over one-unit cases. */
function understoodMention(node: Record<string, unknown>): string | null {
// count() visits every object in the body (types, locations, cases), so
// only the node kinds handled here are read as IR nodes
const kind = node.kind;
if (kind !== "varDecl" && kind !== "strEq" && kind !== "strCmp" && kind !== "switch") return null;
const n = node as unknown as IrExpr | IrStmt;
if (n.kind === "varDecl") return n.init && isCharAt(n.init) ? n.localId : null;
if (n.kind === "strEq" || n.kind === "strCmp") return unitUse(n)?.id ?? null;
@@ -208,6 +208,7 @@ const MODULE_FIELDS: Record<StringFields<Omit<IrModule, "functions">>, Role> = {
"contracts.int32Returns[]": "fact",
"contracts.int32Locals[]": "other",
"contracts.int32Elements[]": "other",
"int32ArrayCandidates[]": "other",
"contracts.noThrow[]": "fact",
"contracts.acyclic[]": "other",
"contracts.immortalGlobals[]": "other",
+515
View File
@@ -0,0 +1,515 @@
import { F64, type IrExpr, type IrFunction, type IrModule, type IrStmt } from "./ir.js";
type SwitchCase = Extract<IrStmt, { kind: "switch" }>["cases"][number];
import { reparseExpr } from "./serialize.js";
import {
everyExprChild,
everyStmtChild,
everyStmtList,
transformExpr,
transformStmt,
transformStmtList,
type IrTransform,
} from "./traverse.js";
/** Local number arrays the backend may store as int32 slots.
*
* A candidate is a local `number[]` of a synchronous function that no
* other binding can hold and only compiled element operations reach:
* every write of the local stores a fresh array (an array literal, or
* `new Array(n).fill(v)`), and every read of it is the receiver of an
* element read, write, state test, `length`, `push` or `pop` (a discarded
* `pop` included), or the array of an element-reader helper call. The
* array never reaches a runtime function that reads its slots, a library
* call, a closure, a return or untyped code.
*
* Candidates are only proposed here. The int32 slot analysis
* (int32-slots.ts) keeps the ones whose every stored element is a proven
* int32 (never -0), assuming the candidates' own elements are int32s, at
* a fixpoint, and the LLVM backend stores those as SCR_ELEM_I32. It keeps
* the ones whose stores are exact integers but not all proven int32s as
* int32 arrays with checked stores: SCR_ELEM_I32 until a store that is not
* an int32 widens them to doubles in place. Their element-reader helper
* calls are inlined first (rewriteCandidates), so no shared helper ever
* receives an int32 array. */
/** Array methods a candidate may be the receiver of: the runtime paths
* behind them read and write int32 slots through the double-valued slot
* accessors. */
export const INT32_ARRAY_METHODS: ReadonlySet<string> = new Set([
"length",
"push",
"pop",
"getNumber",
]);
/** An element-reader helper (`%arr.idxOr.*`): `(a, i) => a's state at i is
* a value ? the element : undefined`, as one returned ternary. */
export function isElementReader(fn: IrFunction): boolean {
if (fn.params.length !== 2 || fn.body.length !== 1) return false;
const [array, index] = fn.params;
const ret = fn.body[0]!;
if (array!.type.kind !== "array" || array!.type.elem.kind !== "f64") return false;
if (index!.type.kind !== "f64" || ret.kind !== "return" || ret.value?.kind !== "ternary")
return false;
const isParam = (e: IrExpr, id: string): boolean => e.kind === "varRef" && e.localId === id;
const { cond, then, else_ } = ret.value;
return (
cond.kind === "bin" &&
cond.op === "===" &&
cond.left.kind === "arrayState" &&
isParam(cond.left.arr, array!.localId) &&
isParam(cond.left.index, index!.localId) &&
cond.right.kind === "numLit" &&
cond.right.value === 1 &&
then.kind === "unionWrap" &&
then.value.kind === "arrayGet" &&
isParam(then.value.arr, array!.localId) &&
isParam(then.value.index, index!.localId) &&
else_.kind === "unionWrap" &&
else_.value.kind === "unitLit"
);
}
/** `new Array(n).fill(v)` as lowering spells it: `fill` over a sequence
* that declares a temporary empty array, sets its length and yields it. */
function fillAllocation(e: IrExpr): (IrExpr & { kind: "arrIntrinsic" }) | null {
if (e.kind !== "arrIntrinsic" || e.method !== "fill" || e.receiver.kind !== "seqExpr")
return null;
const seq = e.receiver;
const [decl, length] = seq.stmts;
if (
seq.stmts.length !== 2 ||
decl?.kind !== "varDecl" ||
decl.init?.kind !== "arrayLit" ||
decl.init.elems.length !== 0 ||
length?.kind !== "arraySetLength" ||
length.arr.kind !== "varRef" ||
length.arr.localId !== decl.localId ||
seq.result.kind !== "varRef" ||
seq.result.localId !== decl.localId
)
return null;
return e;
}
/** Fill-form arrays shorter than this hold doubles even when they are int32
* arrays with checked stores: their doubles fit a typical 32 KiB L1 data
* cache, so 4-byte slots save no memory traffic. */
export const CHECKED_INT32_MIN_LENGTH = 4096;
/** The length `n` of `new Array(n).fill(v)`, or null for any other
* allocation. */
export function fillAllocationLength(e: IrExpr): IrExpr | null {
const fill = fillAllocation(e);
if (!fill || fill.receiver.kind !== "seqExpr") return null;
const set = fill.receiver.stmts[1]!;
return set.kind === "arraySetLength" ? set.length : null;
}
/** A number expression whose evaluation has no effect: locals, literals and
* arithmetic. */
function isPureLength(e: IrExpr): boolean {
if (e.type.kind !== "f64") return false;
switch (e.kind) {
case "numLit":
case "varRef":
return true;
case "bin":
return e.left.type.kind === "f64" && isPureLength(e.left) && isPureLength(e.right);
case "unary":
return isPureLength(e.operand);
default:
return false;
}
}
/** Whether the candidate `id` of `fn` may hold int32 slots with checked
* stores: every array it is assigned is the fill form, with a length that
* may reach CHECKED_INT32_MIN_LENGTH and evaluates without effects (the
* backend evaluates it once more to pick the new array's kind). */
export function checkableInt32Allocations(fn: IrFunction, id: string): boolean {
let ok = true;
everyStmtList(fn.body, {
stmt: (s) => {
if ((s.kind === "varDecl" || s.kind === "assign") && s.localId === id) {
const value = s.kind === "varDecl" ? s.init : s.value;
const length = value ? fillAllocationLength(value) : null;
if (
length === null ||
!isPureLength(length) ||
(length.kind === "numLit" && length.value < CHECKED_INT32_MIN_LENGTH)
)
ok = false;
}
return ok;
},
expr: () => ok,
});
return ok;
}
/** A fresh number array: a literal without spreads, or the fill form. */
export function int32ArrayAllocation(e: IrExpr): boolean {
return (e.kind === "arrayLit" && (e.spreads?.length ?? 0) === 0) || fillAllocation(e) !== null;
}
/** The array literal a candidate's allocation builds (the literal itself,
* or the fill form's temporary's), and the fill form's temporary. */
export function allocatedLiteral(
e: IrExpr,
): { literal: IrExpr & { kind: "arrayLit" }; temporary: string | null } | null {
if (e.kind === "arrayLit") return { literal: e, temporary: null };
const fill = fillAllocation(e);
if (!fill || fill.receiver.kind !== "seqExpr") return null;
const decl = fill.receiver.stmts[0]!;
return decl.kind === "varDecl" && decl.init?.kind === "arrayLit"
? { literal: decl.init, temporary: decl.localId }
: null;
}
/** `xs.pop()` whose value is unused, as lowering spells it: a sequence
* that binds the array to a hidden local, then only reads that local's
* length and truncates it with pop stores. Returns the hidden local and the
* array read it binds. */
function discardedPop(e: IrExpr): { local: string; array: IrExpr & { kind: "varRef" } } | null {
if (e.kind !== "seqExpr") return null;
const [bind, ...rest] = e.stmts;
if (bind?.kind !== "varDecl" || bind.init?.kind !== "varRef") return null;
const local = bind.localId;
const isLocal = (x: IrExpr): boolean => x.kind === "varRef" && x.localId === local;
const expr = (x: IrExpr): boolean => {
if (x.kind === "arrIntrinsic" && x.method === "length" && isLocal(x.receiver)) return true;
return !isLocal(x) && everyExprChild(x, expr, stmt);
};
const stmt = (x: IrStmt): boolean => {
if (x.kind === "arraySetLength" && x.pop === true && isLocal(x.arr)) return expr(x.length);
return (
!((x.kind === "varDecl" || x.kind === "assign") && x.localId === local) &&
everyStmtChild(x, expr, stmt)
);
};
return rest.every(stmt) && expr(e.result) ? { local, array: bind.init } : null;
}
/** Whether evaluating `e` may assign the local `id`. */
function writesLocal(e: IrExpr, id: string): boolean {
const expr = (x: IrExpr): boolean =>
!(x.kind === "assignExpr" && x.localId === id) && everyExprChild(x, expr, stmt);
const stmt = (x: IrStmt): boolean =>
!((x.kind === "varDecl" || x.kind === "assign") && x.localId === id) &&
everyStmtChild(x, expr, stmt);
return !expr(e);
}
/** The candidate locals of `fn` (see the module comment). */
export function int32ArrayCandidates(fn: IrFunction, readers: ReadonlySet<string>): Set<string> {
const result = new Set<string>();
if (fn.async || fn.generator) return result;
const captured = new Set(
[...(fn.captures ?? []), ...(fn.classCaptures ?? [])].map((c) => c.localId),
);
const params = new Set(fn.params.map((p) => p.localId));
for (const local of fn.locals)
if (
local.type.kind === "array" &&
local.type.elem.kind === "f64" &&
!local.boxed &&
!local.tdz &&
!captured.has(local.id) &&
!params.has(local.id)
)
result.add(local.id);
if (result.size === 0) return result;
const written = new Set<string>();
// A receiver position: the candidate is read there without escaping.
const receiver = (e: IrExpr): void => {
if (e.kind !== "varRef") expr(e);
};
const expr = (e: IrExpr): boolean => {
switch (e.kind) {
case "varRef":
result.delete(e.localId);
return true;
case "assignExpr":
result.delete(e.localId);
break;
case "arrayGet":
case "arrayState":
case "arrayHas":
receiver(e.arr);
expr(e.index);
return true;
case "arrIntrinsic":
if (!INT32_ARRAY_METHODS.has(e.method)) break;
receiver(e.receiver);
e.args.forEach(expr);
return true;
case "seqExpr": {
// The pop's hidden local is the array itself; every use of it is a
// length read or a pop store.
const pop = discardedPop(e);
if (!pop) break;
e.stmts.slice(1).every(stmt);
expr(e.result);
return true;
}
case "call": {
if (!readers.has(e.callee) || e.args.length !== 2) break;
const [array, index] = e.args;
// Inlining evaluates the index before the array read.
if (array!.kind === "varRef" && writesLocal(index!, array!.localId))
result.delete(array!.localId);
receiver(array!);
expr(index!);
return true;
}
}
return everyExprChild(e, expr, stmt);
};
const stmt = (s: IrStmt): boolean => {
switch (s.kind) {
case "varDecl":
case "assign": {
const value = s.kind === "varDecl" ? s.init : s.value;
if (value && int32ArrayAllocation(value)) written.add(s.localId);
else if (value) result.delete(s.localId);
if (value) expr(value);
return true;
}
case "arraySet":
receiver(s.arr);
expr(s.index);
expr(s.value);
return true;
case "arraySetUndefined":
// Stores a present undefined: a state, and a cleared (+0) slot.
receiver(s.arr);
expr(s.index);
return true;
case "forOf":
result.delete(s.localId);
break;
case "tryCatch":
if (s.catchLocalId) result.delete(s.catchLocalId);
break;
}
return everyStmtChild(s, expr, stmt);
};
fn.body.every(stmt);
for (const id of result) if (!written.has(id)) result.delete(id);
return result;
}
/** The element values stored into each candidate: element writes, pushed
* values, literal elements and fill values. */
export function int32ArrayStores(
fn: IrFunction,
candidates: ReadonlySet<string>,
): Map<string, IrExpr[]> {
const stores = new Map<string, IrExpr[]>();
const add = (id: string, value: IrExpr): void => {
const list = stores.get(id);
if (list) list.push(value);
else stores.set(id, [value]);
};
const target = (e: IrExpr): string | null =>
e.kind === "varRef" && candidates.has(e.localId) ? e.localId : null;
const expr = (e: IrExpr): boolean => {
if (e.kind === "arrIntrinsic" && e.method === "push") {
const id = target(e.receiver);
if (id !== null) for (const arg of e.args) add(id, arg);
}
return everyExprChild(e, expr, stmt);
};
const stmt = (s: IrStmt): boolean => {
if ((s.kind === "varDecl" || s.kind === "assign") && candidates.has(s.localId)) {
const value = s.kind === "varDecl" ? s.init : s.value;
if (value?.kind === "arrayLit") for (const el of value.elems) add(s.localId, el);
else if (value) {
const fill = fillAllocation(value);
if (fill) add(s.localId, fill.args[0]!);
}
}
if (s.kind === "arraySet") {
const id = target(s.arr);
if (id !== null) add(id, s.value);
}
return everyStmtChild(s, expr, stmt);
};
fn.body.every(stmt);
return stores;
}
/** Rewrite a function's candidates for the backend. A discarded pop uses
* the candidate in place of its hidden local. An element-reader helper call
* whose array is a candidate is inlined: the helper's returned ternary, with
* the candidate for its array and the index for its index, so the read means
* the same. An index that is not a plain read or literal is bound to a fresh
* local first: in its own declaration just before the statement when the
* call is the whole value of a declaration or assignment (the form lowering
* gives `a[i]!`, and the one the backend reads without a box), otherwise in a
* sequence around the ternary. Selection keeps out indexes that write the
* array, so evaluating the index before the array read is unobservable. */
function rewriteCandidates(
fn: IrFunction,
candidates: ReadonlySet<string>,
readers: ReadonlyMap<string, IrFunction>,
): IrFunction {
const locals = [...fn.locals];
let fresh = 0;
const bind = (index: IrExpr): { decl: IrStmt; ref: IrExpr } => {
const id = `%int32.read.${fresh++}`;
locals.push({ id, name: id, type: F64, mutable: false });
return {
decl: { kind: "varDecl", localId: id, init: index, loc: index.loc },
ref: { kind: "varRef", localId: id, type: F64, loc: index.loc },
};
};
const readerCall = (e: IrExpr | null): (IrExpr & { kind: "call" }) | null =>
e?.kind === "call" &&
e.args.length === 2 &&
readers.has(e.callee) &&
e.args[0]!.kind === "varRef" &&
candidates.has(e.args[0]!.localId)
? e
: null;
const simple = (e: IrExpr): boolean => e.kind === "varRef" || e.kind === "numLit";
// The statements of a list, with complex read indexes declared first.
const hoist = (list: IrStmt[]): IrStmt[] =>
list.flatMap((s): IrStmt[] => {
if (s.kind !== "varDecl" && s.kind !== "assign") return [s];
const value = s.kind === "varDecl" ? s.init : s.value;
const call = readerCall(value);
if (!call || simple(call.args[1]!)) return [s];
const { decl, ref } = bind(call.args[1]!);
const read: IrExpr = { ...call, args: [call.args[0]!, ref] };
if (s.kind === "varDecl") return [decl, { ...s, init: read }];
return [decl, { ...s, value: read }];
});
const hoistCase = (c: SwitchCase): SwitchCase => ({ ...c, body: hoist(c.body) });
const lists = (s: IrStmt): IrStmt => {
switch (s.kind) {
case "if":
return { ...s, then: hoist(s.then), else_: s.else_ === null ? null : hoist(s.else_) };
case "block":
return { ...s, body: hoist(s.body) };
case "while":
return { ...s, body: hoist(s.body) };
case "doWhile":
return { ...s, body: hoist(s.body) };
case "for":
return { ...s, body: hoist(s.body) };
case "forOf":
return { ...s, body: hoist(s.body) };
case "switch":
return { ...s, cases: s.cases.map((c) => hoistCase(c)) };
case "tryCatch":
return {
...s,
tryBody: hoist(s.tryBody),
catchBody: s.catchBody === null ? null : hoist(s.catchBody),
finallyBody: s.finallyBody === null ? null : hoist(s.finallyBody),
};
}
return s;
};
// Hidden pop locals replaced by their candidate array.
const popLocals = new Set<string>();
const rewrite = (e: IrExpr): IrExpr => {
const pop = discardedPop(e);
if (e.kind === "seqExpr" && pop && candidates.has(pop.array.localId)) {
popLocals.add(pop.local);
const array = pop.array;
const substitute: IrTransform = {
expr: (x) =>
x.kind === "varRef" && x.localId === pop.local ? { ...array, loc: x.loc } : x,
stmt: (x) => x,
};
return {
...e,
stmts: e.stmts.slice(1).map((x) => transformStmt(x, substitute)),
result: transformExpr(e.result, substitute),
};
}
const call = readerCall(e);
if (!call) {
if (e.kind === "seqExpr") return { ...e, stmts: hoist(e.stmts) };
return e;
}
const helper = readers.get(call.callee)!;
const array = call.args[0]!;
const [arrayParam, indexParam] = helper.params;
const ret = helper.body[0]!;
if (ret.kind !== "return" || !ret.value) return e;
let index = call.args[1]!;
const stmts: IrStmt[] = [];
if (!simple(index)) {
const { decl, ref } = bind(index);
stmts.push(decl);
index = ref;
}
const operand = index;
const body = transformExpr(reparseExpr(ret.value), {
expr: (x) => {
if (x.kind !== "varRef") return x;
if (x.localId === arrayParam!.localId) return reparseExpr(array);
if (x.localId === indexParam!.localId) return reparseExpr(operand);
return x;
},
stmt: (x) => x,
});
// The ternary is a fresh copy (reparseExpr): it takes the call's location.
const typed = body;
typed.loc = e.loc;
return stmts.length === 0
? typed
: { kind: "seqExpr", stmts, result: typed, type: typed.type, loc: e.loc };
};
const body = transformStmtList(hoist(fn.body), { expr: rewrite, stmt: lists });
if (fresh === 0 && popLocals.size === 0) return { ...fn, body };
return { ...fn, body, locals: locals.filter((local) => !popLocals.has(local.id)) };
}
/** Propose the module's int32 array candidates (`function\0local` keys in
* `int32ArrayCandidates`), only those in `keep` when given, and rewrite
* their functions. */
export function markInt32ArrayCandidates(
mod: IrModule,
keep: ReadonlySet<string> | null = null,
): IrModule {
if (mod.lib !== undefined) return mod;
const readers = new Map<string, IrFunction>();
for (const fn of mod.functions) if (isElementReader(fn)) readers.set(fn.name, fn);
const names = new Set(readers.keys());
const keys: string[] = [];
const functions = mod.functions.map((fn) => {
const candidates = int32ArrayCandidates(fn, names);
for (const id of candidates)
if (keep === null || keep.has(`${fn.name}\0${id}`)) keys.push(`${fn.name}\0${id}`);
else candidates.delete(id);
return candidates.size === 0 ? fn : rewriteCandidates(fn, candidates, readers);
});
return keys.length === 0 ? mod : { ...mod, functions, int32ArrayCandidates: keys };
}
/** The module the backend emits: rewritten for exactly the candidates the
* int32 slot analysis (`prove`, returning its int32Arrays) keeps, so a
* module without int32 arrays is `mod` itself. A dropped candidate's
* rewrite is undone by marking again without it, until every candidate
* marked is kept. */
export function selectInt32Arrays(
mod: IrModule,
prove: (marked: IrModule) => ReadonlySet<string>,
): IrModule {
let keep: ReadonlySet<string> | null = null;
for (let round = 0; round < 4; round++) {
const marked = markInt32ArrayCandidates(mod, keep);
if (marked === mod) return mod;
const kept = prove(marked);
const proposed = marked.int32ArrayCandidates ?? [];
if (proposed.every((key) => kept.has(key))) return marked;
keep = new Set(proposed.filter((key) => kept.has(key)));
}
return mod;
}
+74 -1
View File
@@ -10,6 +10,7 @@ import {
} from "./integer-ranges.js";
import { exprFunctionReference, moduleFunctionReferences } from "./function-references.js";
import { findInitializerBindings, withInitializerBindings } from "./initializer-bindings.js";
import { checkableInt32Allocations, int32ArrayStores } from "./int32-arrays.js";
import { everyExprChild, everyStmtChild, everyStmtList, everyTypeChild } from "./traverse.js";
/** The storage slot an array-valued expression reads, as an
@@ -169,7 +170,10 @@ export class Int32Slots {
this.int32Locals.has(`${fnName}\0${localId}`) || this.int32Locals.has(`\0${localId}`)
? INT32_RANGE
: null;
if (this.int32Elements.size > 0) result.elements = (arr) => this.hasInt32Elements(fnName, arr);
if (this.int32Elements.size > 0) {
result.elements = (arr) => this.hasInt32Elements(fnName, arr);
if (this.narrowedReads) result.narrowedReads = true;
}
const params = this.paramFacts.get(fnName);
if (params)
result.params =
@@ -234,6 +238,28 @@ export class Int32Slots {
addInt32Elements(key: string): void {
this.int32Elements.add(key);
}
dropInt32Elements(key: string): boolean {
return this.int32Elements.delete(key);
}
hasInt32ElementSlot(key: string): boolean {
return this.int32Elements.has(key);
}
/** Whether the module has int32 array candidates (the narrowedReads
* range fact). */
narrowedReads = false;
/** Local number arrays (`function\0local`) stored as int32 slots: the
* module's int32 array candidates whose every stored element the
* fixpoint proved an int32. */
readonly int32Arrays = new Set<string>();
/** Local number arrays that may hold int32 slots with checked stores: the
* candidates whose element stores the fixpoint proved exact integers but
* not int32s, allocated only by fills that may be long. A fill of at least
* CHECKED_INT32_MIN_LENGTH elements creates one with int32 slots and a
* shorter one with doubles. Their elements give no int32 facts; compiled
* accesses test the array's kind (once per innermost loop where they can,
* backend/llvm/int32-kind-loops.ts), and a store that is not an int32
* widens it first (scr_arr_widen_i32). */
readonly checkedInt32Arrays = new Set<string>();
/** Whether every present element of the number array `arr` evaluates to
* (read inside `fnName`) is an int32. */
hasInt32Elements(fnName: string, arr: IrExpr): boolean {
@@ -752,6 +778,18 @@ export function analyzeInt32Slots(mod: IrModule, inputs: Int32SlotInputs = {}):
}
for (const key of contracts?.int32Locals ?? []) slots.addInt32Local(key);
for (const key of contracts?.int32Elements ?? []) slots.addInt32Elements(key);
// Int32 array candidates (ir/int32-arrays.ts): assumed to hold int32
// elements until a store the fixpoint cannot prove drops one.
const arrayCandidates = new Map<string, Set<string>>();
slots.narrowedReads = (mod.int32ArrayCandidates?.length ?? 0) > 0;
for (const key of mod.int32ArrayCandidates ?? []) {
const cut = key.indexOf("\0");
const fnName = key.slice(0, cut);
const set = arrayCandidates.get(fnName) ?? new Set<string>();
set.add(key.slice(cut + 1));
arrayCandidates.set(fnName, set);
slots.addInt32Elements(`l:${key}`);
}
/* ── candidate fields ── */
const children = new Map<string, string[]>();
@@ -1031,6 +1069,7 @@ export function analyzeInt32Slots(mod: IrModule, inputs: Int32SlotInputs = {}):
return false;
};
const hasCandidateWrite = (fn: IrFunction): boolean => {
if (arrayCandidates.has(fn.name)) return true;
const box = (localId: string): boolean => slots.hasBox(slots.boxGroup(fn.name, localId));
const returns = slots.hasReturn(fn.name);
return !everyStmtList(fn.body, {
@@ -1076,6 +1115,36 @@ export function analyzeInt32Slots(mod: IrModule, inputs: Int32SlotInputs = {}):
continue;
}
const int32 = (e: IrExpr): boolean => withinInt32(ranges.get(e));
// Every element stored into an int32 array candidate must be an int32
// under the facts that assume the candidates' own elements are; a
// failing store drops its array's element facts, and the function is
// analyzed again without them. When every failing store is still an
// exact integer under those facts (a counter, `a[i]! + 1`) and the array
// is only ever allocated by a fill that may be long
// (checkableInt32Allocations), it keeps int32 slots with checked stores
// (checkedInt32Arrays), which widen it to doubles in place at the first
// store that is not an int32.
const arrays = arrayCandidates.get(fn.name);
if (arrays) {
let dropped = false;
for (const [id, values] of int32ArrayStores(fn, arrays)) {
const key = `l:${fn.name}\0${id}`;
if (!slots.hasInt32ElementSlot(key) || values.every(int32)) continue;
slots.dropInt32Elements(key);
const checkable = (value: IrExpr): boolean => {
const range = ranges.get(value);
return int32(value) || range != null;
};
if (values.every(checkable) && checkableInt32Allocations(fn, id))
slots.checkedInt32Arrays.add(`${fn.name}\0${id}`);
else reason(`array:${fn.name}\0${id}`, "an element store is not int32");
dropped = true;
}
if (dropped && !queued.has(fn.name)) {
queued.add(fn.name);
queue.push(fn.name);
}
}
// An int32 written into a slot still assumed non-negative: widen the
// assumption when the value may be negative, and reanalyze its readers.
const settle = (slot: string, e: IrExpr): void => {
@@ -1144,6 +1213,10 @@ export function analyzeInt32Slots(mod: IrModule, inputs: Int32SlotInputs = {}):
}
for (const proven of provenByFunction.values())
for (const value of proven) slots.provenWrites.add(value);
for (const [fnName, ids] of arrayCandidates)
for (const id of ids)
if (slots.hasInt32ElementSlot(`l:${fnName}\0${id}`))
slots.int32Arrays.add(`${fnName}\0${id}`);
const ranged = rangedParameters(slots, functions, retry, numeric, (callee) =>
readers.get(`ret:${callee}`),
);
+130 -16
View File
@@ -220,6 +220,11 @@ export interface IntegerSlotFacts {
/** Number arrays whose every present element is an int32 (Int32[]
* contracts): an element read that found a value reads an int32. */
elements?: (arr: IrExpr) => boolean;
/** Element reads narrowed out of a union (`xs[i]!`, also when a union
* local holds the read) keep the element's range, and a ternary arm that
* always throws yields no value. Only modules with int32 array candidates
* (ir/int32-arrays.ts) set it, so no other program's ranges change. */
narrowedReads?: boolean;
}
/** A local read, or a numeric local plus or minus an integer literal
@@ -240,6 +245,50 @@ function offsetRef(e: IrExpr): { localId: string; offset: number } | null {
return null;
}
/** Library calls that always throw. */
const THROWING_LIB = /^error\.\w*Throw$/;
function throws(e: IrExpr): boolean {
return e.kind === "libCall" && THROWING_LIB.test(e.fn);
}
/** The values written to each union-typed local of `fn` by declarations
* and assignments; null for a local that another write reaches or that a
* closure captures. */
function unionLocalWrites(
fn: IrFunction,
captures: ReadonlySet<string>,
): Map<string, IrExpr[] | null> {
const unionLocals = new Set(
fn.locals
.filter((l) => l.type.kind === "union" && !l.boxed && !captures.has(l.id))
.map((l) => l.id),
);
const result = new Map<string, IrExpr[] | null>();
const add = (id: string, value: IrExpr | null): void => {
if (!unionLocals.has(id)) return;
const list = result.get(id);
if (list === null) return;
if (value === null) result.set(id, null);
else if (list) list.push(value);
else result.set(id, [value]);
};
everyStmtList(fn.body, {
stmt: (s) => {
if (s.kind === "varDecl" && s.init) add(s.localId, s.init);
else if (s.kind === "assign") add(s.localId, s.value);
else if (s.kind === "forOf") add(s.localId, null);
else if (s.kind === "tryCatch" && s.catchLocalId) add(s.catchLocalId, null);
return true;
},
expr: (e) => {
if (e.kind === "assignExpr") add(e.localId, null);
return true;
},
});
return result;
}
/** Structured local dataflow. Calls cannot write uncaptured local slots,
* but argument expressions can; unknown evaluation forms invalidate all
* syntactically written slots before their children are inspected. Loops
@@ -272,6 +321,10 @@ export function analyzeIntegerRangeFacts(
.map((l) => l.id),
);
const loopLocals = new Map(fn.locals.filter((l) => !captures.has(l.id)).map((l) => [l.id, l]));
// With narrowed reads, the values written to union locals (an element
// read held before it is narrowed), or null for a local some other write
// (assignExpr, for-of, catch, capture) reaches.
const unionWrites = slots.narrowedReads ? unionLocalWrites(fn, captures) : null;
// A captured binding has no local dataflow: only a proof that covers
// every write of the shared box anywhere applies to its reads.
const boxedRange = (localId: string): IntegerRange | null => {
@@ -490,6 +543,45 @@ export function analyzeIntegerRangeFacts(
return range;
}
/** narrowedRange through a union local: every value written to it. */
function narrowedValue(
value: IrExpr,
unionId: string,
tag: number,
): IntegerRange | null | "never" {
if (value.kind !== "varRef") return narrowedRange(value, unionId, tag);
const written = unionWrites?.get(value.localId);
if (!written || written.length === 0) return null;
let result: IntegerRange | null | "never" = "never";
for (const w of written) {
const arm = narrowedRange(w, unionId, tag);
if (arm === "never") continue;
if (arm === null) return null;
result = result === "never" ? arm : union(result, arm);
}
return result;
}
/** The range of the values of union tag `tag` that `value` (already
* analyzed) can hold: "never" when no arm holds that tag. */
function narrowedRange(
value: IrExpr,
unionId: string,
tag: number,
): IntegerRange | null | "never" {
if (value.kind === "seqExpr") return narrowedRange(value.result, unionId, tag);
if (value.kind === "ternary") {
const yes = narrowedRange(value.then, unionId, tag),
no = narrowedRange(value.else_, unionId, tag);
if (yes === "never") return no;
if (no === "never") return yes;
return yes && no ? union(yes, no) : null;
}
if (value.kind === "unionWrap" && value.unionId === unionId) {
if (value.tag !== tag) return "never";
return value.value.kind === "arrayGet" ? (ranges.get(value.value) ?? null) : null;
}
return null;
}
function remember(e: IrExpr, range: IntegerRange | null): IntegerRange | null {
// Shared IR objects have to satisfy the proof at every occurrence.
const previous = ranges.get(e);
@@ -683,6 +775,10 @@ export function analyzeIntegerRangeFacts(
right = expr(e.else_, no);
merge(facts, yes, no);
if (left && right) range = union(left, right);
// With narrowed reads: an arm that always throws yields no value,
// so the other arm's range is the result's.
else if (slots.narrowedReads && throws(e.then)) range = right;
else if (slots.narrowedReads && throws(e.else_)) range = left;
break;
}
case "logical": {
@@ -777,27 +873,45 @@ export function analyzeIntegerRangeFacts(
expr(e.obj, facts);
range = slotRange(e);
break;
case "unionNarrow":
// Only modules with narrowed reads take this rule: the narrow
// evaluates its operand first, and throws for every arm of another
// tag, so an element read `xs[i]!` (a ternary of the wrapped
// element and a wrapped undefined) has the element's range.
if (!slots.narrowedReads || e.type.kind !== "f64") {
range = conservative(e, facts);
break;
}
expr(e.value, facts);
{
const arm = narrowedValue(e.value, e.unionId, e.tag);
range = arm === "never" ? null : arm;
}
break;
default:
// Do not assume the traversal order is an evaluation order for an
// unknown node. Every child gets the same conservative entry state.
havoc(facts, [e], facts);
everyExprChild(
e,
(child) => {
expr(child, new Map(facts));
return true;
},
(child) => {
body([child], new Map(facts));
return true;
},
);
// A direct call's result carries its whole-program return proof.
range = slotRange(e);
range = conservative(e, facts);
break;
}
return remember(e, range);
}
/** An unknown node: do not assume the traversal order is an evaluation
* order. Every child gets the same conservative entry state. */
function conservative(e: IrExpr, facts: Facts): IntegerRange | null {
havoc(facts, [e], facts);
everyExprChild(
e,
(child) => {
expr(child, new Map(facts));
return true;
},
(child) => {
body([child], new Map(facts));
return true;
},
);
// A direct call's result carries its whole-program return proof.
return slotRange(e);
}
function induction(
s: IrStmt & { kind: "for" },
facts: Facts,
+9 -2
View File
@@ -1077,6 +1077,12 @@ export interface IrModule {
* write in the program (ir/contract-check.ts). Backends may rely on them
* unconditionally. Absent when the program writes no contract. */
contracts?: IrContracts;
/** Local number arrays (`function\0local`) the backend may store as
* int32 slots, proposed by ir/int32-arrays.ts after their element-reader
* calls were inlined; the int32 slot analysis keeps the ones whose every
* stored element it proves an int32, and those whose stores are exact
* integers with checked stores. Absent when there is none. */
int32ArrayCandidates?: string[];
/** The WebAssembly modules translated into this program (wasm/
* translate.ts), present exactly when the program uses the WebAssembly
* API (possibly empty). The backend appends each module's translation
@@ -2098,8 +2104,9 @@ export type IrStmt =
* and whose `code` carries the SC diagnostic code; unwinds exactly like
* `throw`.
* TypeScript sources produce it only as the tsc-unreachable fallthrough
* trap (appendImplicitUndefinedReturn's SC9002) — their construct fences
* stay compile errors. */
* trap (appendImplicitUndefinedReturn's SC9002) and for a typed-array
* store through an element index that is undefined at run time
* (undefined-index.ts); their construct fences stay compile errors. */
| { kind: "runtimeFence"; code: string; message: string; loc: SrcLoc }
/** Rethrow of a catch binding (`throw e` where e is the binding):
* re-raises the SAVED exception exactly — kind and payload preserved,
@@ -198,3 +198,24 @@ test("byte decoding keeps the whole-document errors and version refusal", () =>
const sentinel = text.replace('"nan"', '"future"');
expect(() => deserializeModuleBytes(Buffer.from(sentinel))).toThrow("Invalid IR number sentinel");
});
test("file roots respell the source locations under them and nothing else", () => {
const mod = numbers([1, 2]);
const shim = { file: '/opt/scriptc "a"/shims/pkg/index.ts', start: 3, end: 4 };
const other = { file: "/opt/elsewhere/shims/pkg/index.ts", start: 5, end: 6 };
const body = mod.functions[0]!.body as IrStmt[];
body.push(
{ kind: "exprStmt", expr: { kind: "numLit", value: 3, type: F64, loc: shim }, loc: shim },
{ kind: "exprStmt", expr: { kind: "numLit", value: 4, type: F64, loc: other }, loc: other },
);
const roots = [{ from: '/opt/scriptc "a"/shims/', to: "scriptc:shims/" }];
const relocated = deserializeModule(serializeModuleParts(mod, roots).join(""));
const files = (relocated.functions[0]!.body as IrStmt[]).map((stmt) => stmt.loc.file);
expect(files).toEqual([
"numbers.ts",
"numbers.ts",
"scriptc:shims/pkg/index.ts",
"/opt/elsewhere/shims/pkg/index.ts",
]);
expect(serializeModuleParts(mod).join("")).toBe(serializeModule(mod, true));
});
+33 -3
View File
@@ -46,11 +46,41 @@ function encodeFunction(fn: IrFunction): string | undefined {
return ordinaryNumbers(fn) ? JSON.stringify(fn) : JSON.stringify(fn, encodeNumber);
}
/** A directory whose source location files an IR artifact names by a
* prefix of its own: a location file `from + rest` serializes as
* `to + rest`. */
export interface SerializedFileRoot {
from: string;
to: string;
}
/** Respells the source location files under `roots` in serialized parts.
* A location's file is the first member of its JSON object, and the
* escaped spelling of a path prefix is the prefix of the escaped path. */
function relocateFiles(parts: string[], roots: readonly SerializedFileRoot[]): string[] {
const needles = roots.map(({ from, to }) => ({
from: `"file":${JSON.stringify(from).slice(0, -1)}`,
to: `"file":${JSON.stringify(to).slice(0, -1)}`,
}));
return parts.map((part) =>
needles.reduce(
(text, { from, to }) => (text.includes(from) ? text.replaceAll(from, to) : text),
part,
),
);
}
/** The compact encoding as ordered parts: the module header up to the
* function array, one part per function (with separate comma parts), and the
* header's remainder. Concatenating the parts yields exactly
* `serializeModule(mod, true)`; writers stream them to disk in batches. */
export function serializeModuleParts(mod: IrModule): string[] {
* `serializeModule(mod, true)`; writers stream them to disk in batches.
* `fileRoots` respells the source locations in those directories, so
* artifacts do not depend on where the compiler that wrote them is
* installed. */
export function serializeModuleParts(
mod: IrModule,
fileRoots: readonly SerializedFileRoot[] = [],
): string[] {
// A replacer reads through live native views. Serializing the complete
// function array through one view refreshes every function capsule for
// each element, making large compiler artifacts quadratic. The sentinel
@@ -65,7 +95,7 @@ export function serializeModuleParts(mod: IrModule): string[] {
parts.push(encodeFunction(fn) ?? "null");
}
parts.push("]" + header.slice(offset + slot.length));
return parts;
return fileRoots.length === 0 ? parts : relocateFiles(parts, fileRoots);
}
/** Compiler artifacts use compact JSON; API consumers can retain the
+5 -1
View File
@@ -4269,7 +4269,11 @@ function validateFunction(
if (!ok) err(`libCall dyn.objEntryPairs must return a [string, any] tuple array`, e.loc);
return true;
}
if (e.fn === "array.namedKeys" || e.fn === "array.hasNamedKey") {
if (
e.fn === "array.namedKeys" ||
e.fn === "array.hasNamedKey" ||
e.fn === "array.hasProperties"
) {
if (e.args[0]?.type.kind !== "array") {
err(`libCall ${e.fn} arg 0: expected an array, got ${e.args[0]?.type.kind}`, e.loc);
}
+8 -1
View File
@@ -14,7 +14,7 @@ import {
} from "node:fs";
import { join } from "node:path";
import { resolveBuildCacheRoot } from "../backend/cache-root.js";
import { writeTextGeneratedSync } from "./text-file.js";
import { joinedParts, writeTextGeneratedSync } from "./text-file.js";
/** UTF-8 bytes one native compiler string can hold (the runtime's
* SCR_STR_MAX_CAP). Exceeding it traps rather than throwing, so payloads
@@ -28,6 +28,13 @@ export function contentDigest(value: string | Uint8Array): string {
return hash.digest("hex");
}
/** contentDigest of the text the parts concatenate to. */
export function partsDigest(parts: readonly string[]): string {
const hash = createHash("sha256");
for (const part of joinedParts(parts)) hash.update(part);
return hash.digest("hex");
}
/** contentDigest of a file's bytes, read in 1 MiB blocks so a large file
* (a module's LLVM text) is never held whole. */
export function fileDigest(path: string): string {
+4 -4
View File
@@ -362,7 +362,7 @@ export class NativeCompiler {
mkdirSync(inputDirectory);
const llvmInput = outputKind === "llvm" ? stagedOutput : join(inputDirectory, "program.ll");
// In slices: the module text is never encoded whole.
writeTextPartsSync(llvmInput, [llvm]);
writeTextPartsSync(llvmInput, llvm);
timing("llvm-write");
if (outputKind === "llvm") {
renameSync(stagedOutput, output);
@@ -496,7 +496,7 @@ export class NativeCompiler {
renameSync(stagedOutput + ".dSYM", output + ".dSYM");
}
mkdirSync(options.outDir, { recursive: true });
writeTextPartsSync(llvmPath, [llvm]);
writeTextPartsSync(llvmPath, llvm);
let irPath: string | undefined;
if (options.emitIr) {
irPath = join(options.outDir, `${stem}.ir.json`);
@@ -585,7 +585,7 @@ export class NativeCompiler {
const llvm = prepared.llvm;
const llvmInput = join(stage, "program.ll");
// In slices: the module text is never encoded whole.
writeTextPartsSync(llvmInput, [llvm]);
writeTextPartsSync(llvmInput, llvm);
timing("llvm-write");
const features = prepared.features;
const pack = selectNativeRuntimePack(
@@ -648,7 +648,7 @@ export class NativeCompiler {
});
}
timing("native-archive");
writeTextPartsSync(llvmPath, [llvm]);
writeTextPartsSync(llvmPath, llvm);
let irPath: string | undefined;
if (options.emitIr) {
irPath = join(options.outDir, `${stem}.lib.ir.json`);
@@ -11,7 +11,7 @@ import {
import { runNativeTool } from "../backend/native-tools.js";
import type { NativeRuntimeSelection } from "../backend/runtime-pack-native.js";
import type { NativeToolchain } from "./toolchain.js";
import { contentDigest, NativeCache } from "./cache.js";
import { contentDigest, NativeCache, partsDigest } from "./cache.js";
interface InputIdentity {
path: string;
@@ -254,7 +254,7 @@ export class NativeExecutableCache {
export function openNativeExecutableCache(
cache: NativeCache | null,
toolchain: NativeToolchain,
llvm: string,
llvm: readonly string[],
options: CompileRequestOptions,
ffi: FfiProfile | null,
pack: NativeRuntimeSelection,
@@ -299,7 +299,7 @@ export function openNativeExecutableCache(
const key = contentDigest(
JSON.stringify({
schema: 1,
llvm: contentDigest(llvm),
llvm: partsDigest(llvm),
options,
toolchain,
runtimeIdentity: pack.packageText,
+88 -18
View File
@@ -11,11 +11,33 @@ import { NativeCache } from "./cache.js";
import { prepareNativeExecutable } from "./prepare.js";
import type { NativeToolchain } from "./toolchain.js";
/** A prepared input with its LLVM parts joined: a cache hit reads the text
* back as one part. */
function text(input: ReturnType<ReturnType<typeof fixture>["prepare"]>) {
return input.ok ? { ...input, llvm: input.llvm.join("") } : input;
}
const scratch: string[] = [];
afterEach(() => {
for (const path of scratch.splice(0)) rmSync(path, { recursive: true, force: true });
});
function fixtureToolchain(): NativeToolchain {
const target = nativeCodegenTarget()!;
return {
compilerVersion: "test",
target,
helper: target.helper,
helperExecutable: "unused",
helperPackageRoot: "unused",
runtimePackRoot: "unused",
linker: "unused",
linkerArgs: [],
dsymutil: "unused",
ts7Executable: ts7Executable(),
};
}
function fixture() {
const temporary = mkdtempSync(join(tmpdir(), "scriptc-native-frontend-"));
scratch.push(temporary);
@@ -32,19 +54,7 @@ function fixture() {
}),
);
writeFileSync(join(root, "package.json"), '{"type":"module"}');
const target = nativeCodegenTarget()!;
const toolchain: NativeToolchain = {
compilerVersion: "test",
target,
helper: target.helper,
helperExecutable: "unused",
helperPackageRoot: "unused",
runtimePackRoot: "unused",
linker: "unused",
linkerArgs: [],
dsymutil: "unused",
ts7Executable: ts7Executable(),
};
const toolchain = fixtureToolchain();
const cache = new NativeCache(join(temporary, "cache"));
const options = {
outPath: join(root, ".scriptc", "main"),
@@ -68,17 +78,77 @@ test("warm frontend reuse survives output creation and invalidates imported sour
if (!first.ok) return;
mkdirSync(f.options.outDir);
writeFileSync(f.options.outPath, "binary placeholder");
writeFileSync(join(f.options.outDir, "main.ll"), first.llvm);
writeFileSync(join(f.options.outDir, "main.ll"), first.llvm.join(""));
const warm = f.prepare();
expect(warm).toEqual(first);
expect(text(warm)).toEqual(text(first));
expect(f.loads()).toBe(1);
writeFileSync(f.dependency, 'export const value = "second";\n');
const edited = f.prepare();
expect(edited.ok).toBe(true);
if (edited.ok) expect(edited.llvm).not.toBe(first.llvm);
if (edited.ok) expect(edited.llvm.join("")).not.toBe(first.llvm.join(""));
expect(f.loads()).toBe(2);
});
test("comment and whitespace edits reuse the frontend output a fresh build reproduces", () => {
const f = fixture();
const first = f.prepare();
expect(first.ok, JSON.stringify(first)).toBe(true);
if (!first.ok) return;
const commented = '// The exported value.\nexport const value = /* unchanged */ "first";\n';
writeFileSync(f.dependency, commented);
const reused = f.prepare();
expect(f.loads()).toBe(1);
expect(reused.ok).toBe(true);
if (!reused.ok) return;
expect(reused.llvm.join("")).toBe(first.llvm.join(""));
expect(new Map(reused.sources).get(f.dependency)).toBe(commented);
// The reused output is the one the edited sources lower to.
const fresh = prepareNativeExecutable(
f.entry,
f.options,
null,
fixtureToolchain(),
(path, npmStatic, externalTypes) => runFrontend(path, loadProgram, npmStatic, externalTypes),
new NativeCache(join(f.root, "fresh-cache")),
);
expect(fresh.ok && fresh.llvm.join("")).toBe(first.llvm.join(""));
// The refreshed entry still notices a later code edit.
writeFileSync(f.dependency, 'export const value = "second";\n');
const edited = f.prepare();
expect(f.loads()).toBe(2);
expect(edited.ok && edited.llvm.join("")).not.toBe(first.llvm.join(""));
});
test("a comment edit inside source text the output carries is rebuilt", () => {
const f = fixture();
// A failing assert(value) reports its call's source text.
const source = (call: string) =>
`import assert from "node:assert";\nexport const value = "first";\n${call};\n`;
writeFileSync(f.dependency, source("assert(value.length > 0)"));
const first = f.prepare();
expect(first.ok, JSON.stringify(first)).toBe(true);
if (!first.ok) return;
writeFileSync(f.dependency, source("assert(value.length > 0 /* checked */)"));
const edited = f.prepare();
expect(f.loads()).toBe(2);
expect(edited.ok && edited.llvm.join("")).not.toBe(first.llvm.join(""));
});
test("a whitespace edit that moves a reported source line is rebuilt", () => {
const f = fixture();
// node:test reports the file, line and column of each test.
const source = (gap: string) =>
`import { test } from "node:test";\nexport const value = "first";\n${gap}test("value", () => {});\n`;
writeFileSync(f.dependency, source(""));
const first = f.prepare();
expect(first.ok, JSON.stringify(first)).toBe(true);
if (!first.ok) return;
writeFileSync(f.dependency, source("\n"));
const edited = f.prepare();
expect(f.loads()).toBe(2);
expect(edited.ok && edited.llvm.join("")).not.toBe(first.llvm.join(""));
});
test("structurally invalid cached features cause a rebuild even with a valid digest", () => {
const f = fixture();
expect(f.prepare().ok).toBe(true);
@@ -101,13 +171,13 @@ test("serialized IR travels as bounded parts through the frontend cache", () =>
expect(mod.functions.length).toBeGreaterThan(0);
const warm = f.prepare({ emitIr: true });
expect(f.loads()).toBe(1);
expect(warm).toEqual(first);
expect(text(warm)).toEqual(text(first));
const family = join(f.cache.root, "frontend");
const key = readdirSync(family).find((name) => /^[0-9a-f]{64}$/.test(name))!;
const payload = JSON.parse(readFileSync(join(family, key), "utf8"));
expect(payload.schema).toBe("scriptc.native-frontend.v3");
expect(payload.input.llvm).toBe("");
expect(payload.input.ir).toBeNull();
expect(payload.llvm).toBe(first.llvm);
expect(payload.llvm).toBe(first.llvm.join(""));
expect(payload.ir).toEqual(first.ir);
});
+93 -18
View File
@@ -7,7 +7,11 @@ import type {
CompileLibraryOptions,
CompileRequestOptions,
} from "../compile-types.js";
import { emitLlvmModule, type ForkPrimitives, lowRegionTarget } from "../backend/llvm/emitter.js";
import {
emitLlvmModuleParts,
type ForkPrimitives,
lowRegionTarget,
} from "../backend/llvm/emitter.js";
import { inlineRcOption } from "../backend/llvm/shapes.js";
import { executableLinkFeatures } from "../backend/executable-features.js";
import type { NativeLinkFeatures } from "../backend/native-link-info-core.js";
@@ -16,26 +20,32 @@ import type { FfiProfile } from "../ffi/ffi-manifest.js";
import type { FrontendFactory } from "../frontend/pipeline.js";
import {
FrontendInputTracker,
frontendInputsSemanticallyMatch,
frontendInputsStillMatch,
validFrontendInputSnapshot,
type FrontendInputSnapshot,
type FrontendInputExclusions,
trackedReadFile,
} from "../frontend/input-tracker.js";
import { compilerOwnedFileRoots } from "../frontend/dts-paths.js";
import { provenanceSources } from "../frontend/provenance-registry.js";
import { processFork } from "../frontend/process-fork.js";
import { tokenEquivalentSource } from "../frontend/token-equivalence.js";
import { llvmBytes, wtf8Bytes } from "../backend/literals.js";
import { prepareLibrary } from "../library/prepare.js";
import type { LibraryProfile } from "../library/library-profile.js";
import type { IrModule } from "../ir/ir.js";
import { deserializeModuleBytes, serializeModuleParts } from "../ir/serialize.js";
import { contentDigest, NATIVE_STRING_BYTES, type NativeCache } from "./cache.js";
import { batchTextParts, textSlices } from "./text-file.js";
import { batchTextParts, joinedParts, textSlices } from "./text-file.js";
import type { NativeToolchain } from "./toolchain.js";
import { nativeFileIdentity } from "./file-identity.js";
export interface NativeExecutableInput {
ok: true;
llvm: string;
/** The module's LLVM text as parts that concatenate to it
* (emitLlvmModuleParts), never joined: a large program's is 150 MB. */
llvm: readonly string[];
/** Compact serialized IR as bounded parts (`batchTextParts`): compiler-scale
* modules exceed one native string. */
ir: string[] | null;
@@ -71,7 +81,8 @@ function* frontendCacheParts(
input: { ...input, llvm: "", ir: null },
});
yield metadata.slice(0, metadata.length - 1) + ',"llvm":"';
for (const slice of textSlices(input.llvm)) yield JSON.stringify(slice).slice(1, -1);
for (const part of joinedParts(input.llvm))
for (const slice of textSlices(part)) yield JSON.stringify(slice).slice(1, -1);
yield '"';
if (input.ir === null) yield ',"ir":null}';
else {
@@ -111,18 +122,71 @@ function publishFrontendCache(
);
}
/** The cached input when every frontend input still matches, else the
* entry the inputs no longer match, for reuse after equivalent edits. */
function readFrontendCache(
cache: NativeCache,
key: string,
exclusions: FrontendInputExclusions,
): NativeExecutableInput | null {
): { input: NativeExecutableInput } | { stale: FrontendCacheEntry } | null {
// The entry is decoded as one string; larger entries are misses.
const bytes = cache.read("frontend", key, NATIVE_STRING_BYTES);
if (bytes === null) return null;
const cached: unknown = JSON.parse(bytes.toString("utf8"));
if (!validCachedInput(cached) || !frontendInputsStillMatch(cached.probes, exclusions))
if (!validCachedInput(cached)) return null;
if (!frontendInputsStillMatch(cached.probes, exclusions)) return { stale: cached };
return { input: { ...cached.input, llvm: [cached.llvm], ir: cached.ir } };
}
/** The stale entry's input when every changed source is token-equivalent to
* the text the entry was built from (edits to whitespace and to comments
* that cannot affect compilation; see token-equivalence.ts). Equivalent
* sources lower to the same IR apart from source locations, which only dev
* builds' debug information and the serialized IR carry. Diagnostics carry
* locations too, so an entry that reported warnings is rebuilt. A changed
* source whose raw text or line positions the frontend read (call-site text
* in messages, comptime callbacks, test locations) is rebuilt as well:
* frontendInputsSemanticallyMatch refuses its `text` probe. */
function equivalentFrontendInput(
stale: FrontendCacheEntry,
options: CompileRequestOptions,
exclusions: FrontendInputExclusions,
): { input: NativeExecutableInput; probes: FrontendInputSnapshot } | null {
if (
options.optimization === "dev" ||
options.sanitize === true ||
options.emitIr === true ||
stale.ir !== null ||
(stale.input.warnings?.length ?? 0) > 0
)
return null;
return { ...cached.input, llvm: cached.llvm, ir: cached.ir };
const match = frontendInputsSemanticallyMatch(
stale.probes,
new Map(stale.input.sources),
tokenEquivalentSource,
exclusions,
);
if (match === null || match.changed.length === 0) return null;
// Runtime fences and unsupported-construct messages name their source as
// `[<code> at <file>:<line>]`, and an equivalent edit can move the line.
if (
match.changed.some(({ path }) =>
stale.llvm.includes(` at ${llvmBytes(wtf8Bytes(path), false)}:`),
)
)
return null;
return {
input: {
...stale.input,
llvm: [stale.llvm],
ir: null,
sources: stale.input.sources.map(([path, text]) => [
path,
match.currentSources.get(path) ?? text,
]),
},
probes: match.snapshot,
};
}
function implementationIdentity(toolchain: NativeToolchain): string {
@@ -284,10 +348,19 @@ export function prepareNativeExecutable(
.sort(([a], [b]) => a.localeCompare(b)),
}),
);
const input = readFrontendCache(cache, key, exclusions);
if (input !== null) {
const cached = readFrontendCache(cache, key, exclusions);
if (cached !== null && "input" in cached) {
timing("frontend-cache-hit");
return input;
return cached.input;
}
const equivalent =
cached === null || outputKind === "ir"
? null
: equivalentFrontendInput(cached.stale, options, exclusions);
if (equivalent !== null) {
publishFrontendCache(cache, key, equivalent.probes, equivalent.input);
timing("frontend-cache-equivalent");
return equivalent.input;
}
} catch {
key = null;
@@ -305,13 +378,13 @@ export function prepareNativeExecutable(
if (irInput !== "") timing("ir-input");
const ir =
outputKind === "ir" || options.emitIr
? batchTextParts(serializeModuleParts(prepared.mod))
? batchTextParts(serializeModuleParts(prepared.mod, compilerOwnedFileRoots()))
: null;
timing("ir-serialize");
const llvm =
outputKind === "ir"
? ""
: emitLlvmModule(prepared.mod, {
? []
: emitLlvmModuleParts(prepared.mod, {
targetTriple: toolchain.target.llvmTriple,
pointerBits: toolchain.target.pointerBits,
wasi: toolchain.target.platform === "wasi",
@@ -399,10 +472,10 @@ export function prepareNativeLibrary(
.sort(([a], [b]) => a.localeCompare(b)),
}),
);
const input = readFrontendCache(cache, key, exclusions);
if (input !== null) {
const cached = readFrontendCache(cache, key, exclusions);
if (cached !== null && "input" in cached) {
timing("frontend-cache-hit");
return input;
return cached.input;
}
} catch {
key = null;
@@ -427,7 +500,7 @@ export function prepareNativeLibrary(
);
});
if (!prepared.ok) return prepared;
const llvm = emitLlvmModule(prepared.mod, {
const llvm = emitLlvmModuleParts(prepared.mod, {
targetTriple: toolchain.target.llvmTriple,
pointerBits: toolchain.target.pointerBits,
wasi: toolchain.target.platform === "wasi",
@@ -438,7 +511,9 @@ export function prepareNativeLibrary(
const input: NativeExecutableInput = {
ok: true,
llvm,
ir: options.emitIr ? batchTextParts(serializeModuleParts(prepared.mod)) : null,
ir: options.emitIr
? batchTextParts(serializeModuleParts(prepared.mod, compilerOwnedFileRoots()))
: null,
sidecarJson: prepared.sidecarJson,
features: executableLinkFeatures(prepared.mod),
sources: [...prepared.sourceTexts],
+11 -1
View File
@@ -2,7 +2,7 @@ import { mkdtempSync, readFileSync, rmSync } from "node:fs";
import { tmpdir } from "node:os";
import { join } from "node:path";
import { afterEach, expect, test } from "vitest";
import { textSlices, writeTextPartsSync } from "./text-file.js";
import { joinedParts, textSlices, writeTextPartsSync } from "./text-file.js";
const directories: string[] = [];
afterEach(() => {
@@ -40,3 +40,13 @@ test("a part larger than one slice is written as its UTF-8 bytes", () => {
expect(seen.join("")).toBe("head," + large + "tail");
expect(Math.max(...seen.map((slice) => slice.length))).toBeLessThanOrEqual(1024 * 1024);
});
test("small parts join into batches and a long part passes through whole", () => {
const small = Array.from({ length: 3000 }, (_, index) => `line ${index}\n`);
const long = "x".repeat(2 * 1024 * 1024);
const parts = [...small, long, "tail"];
const joined = [...joinedParts(parts)];
expect(joined.join("")).toBe(parts.join(""));
expect(joined.length).toBeLessThan(5);
expect(joined).toContain("tail");
});
+21 -3
View File
@@ -42,6 +42,24 @@ export function* textSlices(text: string, length = TEXT_WRITE_LENGTH): Generator
}
}
/** `parts` with consecutive parts joined into strings of about
* TEXT_WRITE_LENGTH UTF-16 units (a longer part passes through whole), so
* writing or escaping many small parts takes few calls. */
export function* joinedParts(parts: readonly string[]): Generator<string> {
let batch: string[] = [];
let length = 0;
for (const part of parts) {
batch.push(part);
length += part.length;
if (length >= TEXT_WRITE_LENGTH) {
yield batch.length === 1 ? batch[0]! : batch.join("");
batch = [];
length = 0;
}
}
if (batch.length > 0) yield batch.join("");
}
/** Encodes and writes `part` to `fd` in slices (`textSlices`), so a large
* part is never encoded whole. `written` sees each slice after it is
* written. */
@@ -55,8 +73,8 @@ function writeTextPart(fd: number, part: string, written?: (slice: string) => vo
}
}
/** Write text parts to `path` in order without joining them first; pass
* batched parts (`batchTextParts`) for many small ones. */
/** Write text parts to `path` in order without joining them whole: small
* parts are written in batches (`joinedParts`). */
export function writeTextPartsSync(
path: string,
parts: readonly string[],
@@ -64,7 +82,7 @@ export function writeTextPartsSync(
): void {
const fd = openSync(path, "w");
try {
for (const part of parts) writeTextPart(fd, part, written);
for (const part of joinedParts(parts)) writeTextPart(fd, part, written);
} finally {
closeSync(fd);
}
@@ -99,11 +99,11 @@ test("native toolchain assets resolve from the physical manifest through package
const direct = loadNativeToolchain(path);
const viaLink = loadNativeToolchain(join(linked, "bin", "scriptc.exe.json"));
expect(viaLink).toEqual(direct);
expect(viaLink.ts7Executable).toBe(join(assets, "typescript/lib/tsc.exe"));
expect(viaLink.ts7Executable).toBe(join(assets, "tsc-ts/lib/tsc-ts.exe"));
expect(
loadNativeToolchain(join(linked, "bin", "scriptc.exe.json"), { SCRIPTC_CHECKER: "tsc-ts" })
loadNativeToolchain(join(linked, "bin", "scriptc.exe.json"), { SCRIPTC_CHECKER: "tsgo" })
.ts7Executable,
).toBe(join(assets, "tsc-ts/lib/tsc-ts.exe"));
).toBe(join(assets, "typescript/lib/tsc.exe"));
expect(viaLink.helperExecutable).toBe(join(assets, "llvm/bin/scriptc-llvm-codegen.exe"));
expect(
loadNativeToolchain(join(linked, "bin", "scriptc.exe.json"), {
+4 -4
View File
@@ -177,11 +177,11 @@ export function loadNativeToolchain(
: [...target.defaultLinkerArgs];
return {
compilerVersion: manifest.compiler_version,
// tsc-ts, the default checker except on Windows, comes with the distribution;
// a development build takes it from the tsc-ts package installed with the
// compiler or for the working directory. SCRIPTC_CHECKER selects a checker.
// tsc-ts, the default checker, comes with the distribution; a development
// build takes it from the tsc-ts package installed with the compiler or for
// the working directory. SCRIPTC_CHECKER selects a checker.
ts7Executable:
checkerBackend(env, host.platform) === "tsgo"
checkerBackend(env) === "tsgo"
? pathFrom(root, manifest.ts7)
: manifest.tsc_ts !== undefined
? pathFrom(root, manifest.tsc_ts)
@@ -0,0 +1,55 @@
import { mkdtemp, readFile, rm, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import { join } from "node:path";
import { expect, test } from "vitest";
import { compile } from "../src/index.js";
async function emitLlvm(source: string): Promise<string> {
const dir = await mkdtemp(join(tmpdir(), "scriptc-index-product-"));
try {
const entry = join(dir, "main.ts");
const outPath = join(dir, "main.ll");
await writeFile(entry, source);
const result = await compile(entry, { outDir: dir, outPath, outputKind: "llvm" });
expect(result.ok, JSON.stringify(result.diagnostics)).toBe(true);
return await readFile(outPath, "utf8");
} finally {
await rm(dir, { recursive: true, force: true });
}
}
const source = `
function at(a: number[], x: number, y: number, k: number): number | undefined {
return a[x * y + k];
}
function cancel(a: number[], p: number, q: number, r: number, s: number): number | undefined {
return a[p * q - r * s];
}
const a = [1, 2, 3];
const v = process.argv.length;
console.log(at(a, v | 0, (v + 1) | 0, -v | 0), cancel(a, v | 0, v | 0, v | 0, v | 0));
`;
function body(llvm: string, name: string): string {
return new RegExp(`^define [^@]*@sc_\\w+?_${name}(\\.i32)?\\([^]*?^}`, "m").exec(llvm)?.[0] ?? "";
}
test("an int32 product plus an int32 indexes as an exact i64 without a round trip", async () => {
const llvm = await emitLlvm(source);
// The int32 entry: the index is the i64 product and sum, compared with
// the array's counts; the double goes only to the slow path's runtime call.
const at = /^define [^@]*@sc_bf_at\.i32\([^]*?^}/m.exec(llvm)?.[0] ?? "";
const product = /(%\S+) = mul nsw i64 %\S+, %\S+/.exec(at)?.[1];
expect(product).toBeDefined();
expect(at).toContain(`= add nsw i64 ${product}, `);
expect(at).not.toMatch(/fptosi double %\S+ to i64/);
});
test("a difference of two int32 products keeps the double's round trip", async () => {
const llvm = await emitLlvm(source);
// Each product may pass 2^53, where its double rounds before the
// difference: the index keeps the exactness test of the double.
const cancel = body(llvm, "cancel");
expect(cancel).toMatch(/fptosi double %\S+ to i64/);
expect(cancel).toMatch(/fcmp oeq double/);
});
@@ -0,0 +1,198 @@
import { mkdtemp, readFile, rm, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import { join } from "node:path";
import { expect, test } from "vitest";
import { compile, deserializeModule } from "../src/index.js";
import type { IrModule } from "../src/ir/ir.js";
import { markInt32ArrayCandidates, selectInt32Arrays } from "../src/ir/int32-arrays.js";
async function emitLlvm(source: string): Promise<string> {
const dir = await mkdtemp(join(tmpdir(), "scriptc-int32-elements-"));
try {
const entry = join(dir, "main.ts");
const outPath = join(dir, "main.ll");
await writeFile(entry, source);
const result = await compile(entry, { outDir: dir, outPath, outputKind: "llvm" });
expect(result.ok, JSON.stringify(result.diagnostics)).toBe(true);
return await readFile(outPath, "utf8");
} finally {
await rm(dir, { recursive: true, force: true });
}
}
async function lowerIr(source: string): Promise<IrModule> {
const dir = await mkdtemp(join(tmpdir(), "scriptc-int32-elements-"));
try {
const entry = join(dir, "main.ts");
const outPath = join(dir, "main.ir.json");
await writeFile(entry, source);
const result = await compile(entry, { outDir: dir, outPath, outputKind: "ir" });
expect(result.ok, JSON.stringify(result.diagnostics)).toBe(true);
return deserializeModule(await readFile(outPath, "utf8"));
} finally {
await rm(dir, { recursive: true, force: true });
}
}
const source = `
function compose(n: number, rounds: number): number {
const p: number[] = [];
const q: number[] = [];
for (let i = 0; i < n; i++) p.push((i * 7919 + 3) & (n - 1));
for (let i = 0; i < n; i++) q.push(i);
let check = 0;
for (let r = 0; r < rounds; r++) {
for (let i = 0; i < n; i++) q[i] = p[q[i]!]!;
check = (check + q[r % n]!) | 0;
}
return check;
}
function counts(n: number): number {
const c = new Array(n).fill(0) as number[];
for (let i = 0; i < n; i++) c[(i * 5) % n] = i;
while (c.length > 1) c.pop();
return c[0]!;
}
function fraction(n: number): number {
const a = [1, 2, 3];
a[1] = n / 3;
return a[1]!;
}
function negativeZero(): number {
const a = [1, -0];
return 1 / a[1]!;
}
function counter(n: number): number {
const a = new Array(n).fill(0) as number[];
for (let i = 0; i < 1000; i++) a[i % n] = a[i % n]! + 1;
return a[0]!;
}
function shortCounter(): number {
const a = new Array(16).fill(0) as number[];
for (let i = 0; i < 1000; i++) a[i % 16] = a[i % 16]! + 1;
return a[0]!;
}
function literalCounter(n: number): number {
const a = [0];
for (let i = 0; i < n; i++) a[0] = a[0]! + 1;
return a[0]!;
}
function joined(n: number): string {
const a = [n | 0, 2];
return a.join(",");
}
const { n } = JSON.parse('{"n": 8}') as { n: number };
console.log(
compose(n | 0, 3),
counts(n | 0),
fraction(n),
negativeZero(),
counter(n),
shortCounter(),
literalCounter(n),
joined(n),
);
`;
/** Every function emitted for the source function \`name\` (its int32 ABI
* body and its wrapper included). */
function functions(llvm: string, name: string): string {
const pattern = new RegExp(`^define [^@]*@sc_b?f_${name}[.\\w]*\\([^]*?^}`, "gm");
return [...llvm.matchAll(pattern)].map((m) => m[0]).join("\n");
}
test("arrays storing only int32 elements get 4-byte slots", async () => {
const llvm = await emitLlvm(source);
const compose = functions(llvm, "compose");
expect(compose).not.toBe("");
// Both arrays are allocated as SCR_ELEM_I32 (7).
expect(compose.match(/@scr_arr_new_small\(i32 7\)/g)?.length).toBe(2);
// Reads load an i32 slot; writes store one.
expect(compose).toMatch(/getelementptr inbounds i32, ptr %\w+, i64 %\w+\n\s+%\w+ = load i32/);
expect(compose).toMatch(/store i32 %\w+, ptr %\w+, !tbaa/);
// The element read through the helper's ternary stays in a stack box.
expect(compose).not.toContain("@scr_union_new_f64");
expect(compose).not.toMatch(/call [^\n]*@sc_b?f__x25_arr_idxOr/);
// The fill form and a discarded pop.
const counts = functions(llvm, "counts");
expect(counts).toMatch(/@scr_arr_new\(i32 7, i64 0\)/);
expect(counts).not.toContain("popArray");
});
test("arrays with a store that has no exact integer range, or that escape, keep doubles", async () => {
const llvm = await emitLlvm(source);
for (const name of ["fraction", "negativeZero", "joined"]) {
const fn = functions(llvm, name);
expect(fn, name).not.toBe("");
expect(fn, name).not.toMatch(/@scr_arr_new(_small)?\(i32 7/);
expect(fn, name).not.toContain("arr.kind.");
}
});
test("a counter array keeps int32 slots with checked stores", async () => {
const llvm = await emitLlvm(source);
const counter = functions(llvm, "counter");
// `a[k] = a[k]! + 1` is an exact integer but no int32 proof: a fill of at
// least 4096 elements creates SCR_ELEM_I32 storage (SCR_ELEM_F64 for a
// shorter one), accesses test the kind, and a store widens the array
// before a value that is not an int32.
expect(counter).toMatch(
/fcmp oge double %\w+, [^\n]+\n\s+%\w+ = select i1 %\w+, i32 7, i32 0\n\s+%\w+ = call ptr @scr_arr_new(_small)?\(i32 %\w+/,
);
expect(counter).toMatch(/load i8, ptr %\w+[^\n]*\n\s+%\w+ = icmp eq i8 %\w+, 7/);
expect(counter).toContain("@llvm.fptosi.sat.i32.f64");
expect(counter).toContain("call void @scr_arr_widen_i32(");
// Arrays with proven stores take no kind test, and a fill of a constant
// short length or an array literal keeps doubles.
for (const name of ["compose", "counts", "shortCounter", "literalCounter"]) {
const fn = functions(llvm, name);
expect(fn, name).not.toContain("arr.kind.");
expect(fn, name).not.toContain("kind.fast");
expect(fn, name).not.toContain("scr_arr_widen_i32");
}
for (const name of ["shortCounter", "literalCounter"])
expect(functions(llvm, name), name).not.toMatch(/@scr_arr_new(_small)?\(i32 7/);
});
test("a loop over a counter array tests its kind once on entry", async () => {
const llvm = await emitLlvm(source);
const counter = functions(llvm, "counter");
// The loop's fast copy runs while the array holds int32 slots, with no
// kind test of its own: the sum is an integer, and a sum past the int32
// range leaves for the third copy, which re-executes the store with kind
// tests and widens the array.
expect(counter).toMatch(
/icmp eq i8 %\w+, 7\n\s+br i1 %\w+, label %kind\.fast\d+, label %kind\.wide/,
);
const rest = counter.slice(counter.search(/^kind\.fast\d+:/m));
// The fast copy is the last one, before the loop's exit.
const fast = rest.slice(0, rest.search(/^loop\.e\d+:/m));
expect(fast).not.toBe("");
expect(fast).not.toMatch(/icmp eq i8 %\w+, 7/);
expect(fast).toMatch(/add nsw i64 %\w+, 1/);
expect(fast).toMatch(
/icmp sle i64 %\w+, 2147483647\n[^\n]*\n\s+br i1 %\w+, label %arr\.store\.i32\d+, label %kind\.resume\d+/,
);
expect(fast).toMatch(/store i32 %\w+, ptr %\w+, !tbaa/);
expect(counter).toMatch(/^kind\.resume\d+:/m);
});
test("only the candidates the proof keeps are rewritten", async () => {
const mod = await lowerIr(source);
const proposed = markInt32ArrayCandidates(mod).int32ArrayCandidates ?? [];
expect(proposed).toEqual(
expect.arrayContaining(["compose\0p.0", "compose\0q.0", "counts\0c.0", "counter\0a.0"]),
);
// An array that reaches `join` is never proposed.
expect(proposed.some((key) => key.startsWith("joined\0"))).toBe(false);
// Without a kept candidate, the module is the input itself.
expect(selectInt32Arrays(mod, () => new Set())).toBe(mod);
const kept = new Set(["compose\0p.0", "compose\0q.0"]);
const selected = selectInt32Arrays(
mod,
(marked) => new Set((marked.int32ArrayCandidates ?? []).filter((key) => kept.has(key))),
);
expect(selected.int32ArrayCandidates).toEqual([...kept]);
for (const fn of selected.functions)
if (fn.name !== "compose") expect(fn).toBe(mod.functions.find((f) => f.name === fn.name));
});
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7037-js-var-open-writes/main.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7038-js-var-later-writes.mjs"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7060-array-numeric-keys.cjs"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7061-array-object-keys.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7062-inspect-array-properties.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7325-array-index-products-near-2-53.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7326-int32-element-arrays.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7327-int32-array-widening.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7364-new-target-late-instantiations.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7402-unknown-array-methods.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7441-untyped-receiver-own-string-method-names.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7480-builtin-constructor-in.js"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7480-class-call-without-new.ts"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7481-class-call-without-new.mjs"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7481-class-runtime-heritage-object-error.js"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7482-define-property-function-class-values.js"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7483-constructor-factory-traits.js"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7484-hoisted-function-captures-later-class.js"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7485-global-object-own-properties.js"
],
"diags": []
}
@@ -0,0 +1,6 @@
{
"order": [
"<repo>/tests/corpus/7486-prototype-accessor-string-method-names.js"
],
"diags": []
}
@@ -0,0 +1,9 @@
{
"order": [
"<repo>/tests/corpus/7487-namespace-spread-untyped/shapes.js",
"<repo>/tests/corpus/7487-namespace-spread-untyped/extra.js",
"<repo>/tests/corpus/7487-namespace-spread-untyped/all.js",
"<repo>/tests/corpus/7487-namespace-spread-untyped/main.js"
],
"diags": []
}

Some files were not shown because too many files have changed in this diff Show More