Commit Graph
18 Commits
Author SHA1 Message Date
Chris Tate b16c739d34 Classify zero-cost gate differences instead of failing every runtime change (#2271) 2026-10-27 19:27:25 -05:00
Chris Tate 08fbecb7e7 Excuse only reproduced differences as nondeterministic, and read vendor archive symbols
A program is nondeterministic only when every artifact that differs also
differs between two baseline builds; a deterministic difference in another
artifact still fails the gate. The runtime-only check now reads the symbols
of changed vendor archives as well as runtime objects.
2026-10-27 17:55:44 -05:00
Malte UblandClaude Opus 5.5 4d5127944e Merge main into the zero-cost triage branch
Main moved the per-program report lines into zero-cost-compare.mjs and
added sharding and a baseline cache. The runtime-only and
nondeterministic verdicts now print through programLines, count as
compared in merged shard reports, and the runtime pack diff is skipped
when only recording a baseline cache.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-27 05:38:08 -07:00
Chris Tate df77671e33 Merge main 2026-10-26 23:13:34 -05:00
Malte UblandClaude Opus 5.5 b4ba4d38ac Key the gate's baseline cache on the linker clang finds
Linux links use ld.lld when clang finds it and the system linker
otherwise, and the two produce different executables. A baseline record
made with one must not settle candidates linked with the other.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-26 19:43:02 -07:00
Chris Tate cd606e96ff Key baseline cache entries by their program's inputs
A recorded baseline entry now holds a hash of its program's inputs: the
entry file and its relative imports (paths and contents) and the
optimization level. --baseline-cache uses an entry only when the program
still hashes the same, so a program or an import that changed since the
record builds its baseline, instead of a recorded build error or artifact
hash of the old sources settling the comparison as both-fail or identical.
Records take schema 2; older records are ignored.
2026-10-26 19:53:49 -05:00
Malte UblandClaude Opus 5.5 d0047c813f Shard the zero-cost gate and reuse baseline work across runs
The full gate takes 30 to 75 minutes on one Sandbox, which made it the
longest step of a validation round. It can now be split and merged:

- --shard=i/n splits the bench and corpus programs (the self-hosted
  compiler runs in shard 1, or alone with --set=self), and
  scripts/zero-cost-merge.mjs merges the shards' --json reports into one
  verdict, recomputing the compile-time intervals over every shard.
- --record-baseline-cache records the baseline's artifact hashes once;
  --baseline-cache skips the baseline build of every program whose
  candidate artifacts hash the same. The record is tied to the built
  baseline checkout, toolchain and paths.
- The candidate-source self build overlaps the comparison builds on
  hosts with room for it.

pnpm test:sandbox can reuse a prepared snapshot's LLVM helper and runtime
pack (SCRIPTC_SANDBOX_NATIVE_REFERENCE) when their inputs are identical,
and retries rate-limited Sandbox creation with backoff.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-26 17:43:44 -07:00
Malte Ubl 5817be4d0f Merge remote-tracking branch 'origin/main' into ni/1-libnode-loader
# Conflicts:
#	packages/compiler/src/diagnostics/diagnostic.ts
#	packages/compiler/src/ir/function-references.test.ts
2026-10-26 11:08:44 -07:00
Malte UblandClaude Opus 5.5 5aa6699b3d Build the candidate's own compiler source in the zero-cost gate
The self program compiles the compiler source of the checkout the programs
come from (the baseline by default), so a change adding compiler code the
self-hosted frontend cannot lower passed the gate and failed CI's
self-hosting tests. With the self set, the gate now also builds the
candidate's packages/compiler/src/native/cli.ts with the candidate (LLVM
only, not compared) and fails when that build fails.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-26 07:09:02 -07:00
Malte UblandClaude Opus 5.5 0d27008899 Classify zero-cost gate differences instead of failing every runtime change
Any change to the runtime's C sources makes every executable differ, so
the gate failed even when every program's IR and LLVM were byte-identical.
The gate now classifies each difference outside --allow:

- runtime-only: IR and LLVM identical, the same LLVM helper, runtime pack
  objects that differ between the checkouts (bitcode alone records the
  checkout path, so it does not count), and every symbol the executable
  gains, loses or resizes defined by a changed runtime object. It passes
  with a note that lists the changed runtime functions the program links
  (from the release and speed bitcode, through llvm-dis) and the size
  change.
- nondeterministic: the baseline builds the program again, and its two
  builds differ in the same artifacts. It passes with a note.
- differs: anything else; it fails as before.

--strict fails runtime-only and nondeterministic programs too.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-25 15:26:18 -07:00
Malte UblandClaude Opus 5.5 9a0a3f1562 Load Node.js in-process on first use: libnode bridge and runtime loader
Scope-B milestone NI-1, runtime side. Programs whose IR has a
`nodeIsland.*` operation link scr_node_island.c and the SCR_NODE_ISLANDS
build of scr_async.c; every other program links neither, byte for byte.

- scr_node_island.c dlopens the bridge of a libnode pack on the first call
  (RTLD_NOW | RTLD_GLOBAL), checks its bridge ABI and that it links exactly
  the Node version recorded at build time, starts one Node environment and
  runs the embedded bootstrap (scr_node_island_host.cjs). A run that never
  calls into Node never maps libnode. The pack directory is SCRIPTC_LIBNODE
  at run time, else the recorded path; a `$ORIGIN/` path resolves against
  the executable's directory (the beside packaging, and the hook for
  shipping Node with the program).
- scr_node_bridge.h is the C ABI between the runtime and the bridge; the
  runtime includes no Node or V8 header. The bridge
  (packages/libnode-common/bridge/scr_node_bridge.cc) initializes Node once
  with no default signal handling, no debug-signal handler or inspector,
  no stdio reinitialization and one platform worker, creates the
  environment from the built-in snapshot, routes Node's process exit to
  the runtime's exit sequence, promotes libnode to the global scope for
  native addons, and runs calls in a CallbackScope with a TryCatch.
  build-pack.mjs assembles a pack (libnode, the bridge, manifest.json) from
  a Node --shared build and the matching headers.
- The SCR_NODE_ISLANDS build of scr_async.c adds a loop slot through which
  the runtime's loop drives Node's until NI-3 hosts the runtime's queues on
  Node's libuv: it keeps the program alive while Node has ref'd work,
  sleeps on Node's backend fd up to its next timer, and runs one
  non-blocking turn per loop turn.
- The `nodeIsland` runtime feature (rule, link features, unit table, pack
  matrix row and scr_async.c variant), the nodeIsland.call/callSync
  libCalls, the install call in main, and the build-time record of the
  pack (node-islands/libnode.ts: --libnode=<pack>|beside or
  SCRIPTC_LIBNODE, SC8005 when none is usable, Node >=24.13.1).
- The zero-cost gate and the bench nm test require that no good-code
  executable defines a Node island symbol.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-21 08:54:47 -07:00
Chris Tate cb75c51eda Fail the zero-cost gate when it compares nothing
A benchmark workload or the self-hosted compiler that fails to build
with both checkouts now fails the gate, and so does a run in which no
program built with both. Before this, a run whose builds all failed the
same way, such as with a helper of another version, reported PASS. A
build error now prints in full under its program; its first line is
often a warning.
2026-10-21 03:45:36 -05:00
Chris Tate fd34d72e00 Keep each runtime benchmark workload in its own file
workloads.json listed every workload in one array, and its README table
listed them in one hand order, so changes that each added a workload
appended at the same place and conflicted. Each workload is now a JSON
file in benchmarks/runtime/workloads/ named after it, read through
benchmarks/runtime/workloads.mjs by the bench runner, the zero-cost
gate, the strace script and the dynamic-tier test, ordered by category
and then name. The loader still reads workloads.json in a checkout from
before this change, so the gate can compare against one. The README
table follows the same order.
2026-10-20 18:08:10 -05:00
Malte UblandClaude Opus 5.5 7ea35785ee Check the bench workloads' dynamic-tier symbols in CI
The zero-cost gate's I3 check (a fully typed bench workload defines no
dynamic-tier runtime symbol) runs only when someone runs the gate, and on
Darwin it never matched: Mach-O names carry a leading underscore, so no
symbol had a DYNAMIC_TIER_PREFIXES prefix.

DYNAMIC_WORKLOADS and the nm function-name reader move to
zero-cost-compare.mjs, which now drops the Mach-O underscore. A harness
test builds every bench program on Linux and fails when a workload outside
DYNAMIC_WORKLOADS links a dynamic-tier symbol, or when a listed workload
no longer links any, so the list shrinks as leaks close.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-19 06:44:29 -07:00
Chris Tate 15636ecf90 List bench/thread-slots among the zero-cost gate's dynamic workloads
The workload's release binary links dynamic-tier symbols through its
worker messages: node:worker_threads values (Worker, parentPort,
workerData) are dyn objects, and worker.on and postMessage lower to
dynamic method calls. ThreadSlots and InternTable link none on their
own. intern-table and typed-array-scan-workers are listed for the same
reason.
2026-10-18 22:18:10 -05:00
Chris Tate 29c8ad237c Throw when nm cannot list an executable's symbols in the zero-cost gate
The gate read an empty symbol list when nm could not start, was killed by
a signal, or exited with a non-zero status, so the dynamic-tier symbol
check and the executable comparison passed without any symbols. nmOutput
now throws with the binary, the reason and nm's stderr.
2026-10-18 11:36:13 -05:00
Malte UblandClaude Opus 5.5 4128e6f6f5 Count instructions for differing corpus executables in the zero-cost gate
--cachegrind now also runs differing corpus executables (without
arguments, accepting intentional non-zero exits), so emitter changes that
only touch corpus programs get per-function instruction deltas. AGENTS.md
mentions the dynamic-tier symbol rule.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-18 08:57:34 -07:00
Malte UblandClaude Opus 5.5 8e6986c561 Add the zero-cost gate
scripts/zero-cost-gate.mjs compiles a fixed good-code set (the runtime
benchmark workloads, the self-hosted compiler, and the TypeScript corpus)
with a baseline and a candidate checkout. It requires byte-identical IR,
LLVM and release executables, and names the first differing IR function,
LLVM function or executable symbol when something changes. Programs that
match the change's pattern signature (--allow) may differ. It also checks
compile time (paired builds, bootstrap interval), binary size, optional
per-function cachegrind instruction counts, and the two-tier I3 symbol
rule: no executable gains dynamic-tier runtime symbols, and fully typed
benchmark workloads define none.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-18 08:17:51 -07:00