- Remove C emission, debugging options, and automatic backend fallback.
- Complete LLVM lowering and native bootstrap coverage across supported targets.
- Link executables and libraries with precompiled C runtime packs, including iOS and Android.
- Implement the Zig 0.16.0 AArch64 musl runtime and preserve compiler failure diagnostics.
- Promote the target through executable, library, localization, and documentation matrices.
- Verify both backends and the complete arm64 Alpine differential contract.
Co-authored-by: Arth Tyagi <41021374+arthtyagi@users.noreply.github.com>
* feat: add mobile library targets for iOS and Android
- three library-mode-only triples join SCRIPTC_TARGET:
aarch64-apple-ios, aarch64-apple-ios-simulator (darwin hosts with
Xcode's iPhoneOS/iPhoneSimulator SDKs), and aarch64-linux-android
(any host with an NDK). The consuming pattern is an embedding app
linking profile-configured static archives — Xcode on the Apple side,
Gradle/NDK on the Android side — so the executable lane refuses the
triples with SC3002 and the pointer to --lib
- zig bundles no Apple or bionic libc: resolveCc discovers the sysroot
(xcrun for the selected Apple SDK; ANDROID_NDK_ROOT or the newest
ndk/<version> under the SDK root) and spells it into targetArgs,
where every cache tier already keys it. The canonical LLVM spellings
map to zig's own with the minimum-version floors pinned:
aarch64-ios.15.0[-simulator] stamps LC_BUILD_VERSION minos 15.0 into
every object, aarch64-linux-android.26 pins the bionic stub level an
embedder's minSdkVersion must meet. API 26 bionic carries everything
the library-lane units call (arc4random_buf included), so no shim TU
joins the archive
- localization follows the existing object-format rule: both iOS
platforms ride the Mach-O host-ld64 arm (ld64 reads iOS objects and
preserves platform/minos through the -r merge), Android rides the
cross-ELF arm (zig relocatable merge + in-process demotion) via the
driver's zig spelling. abi.instance_per_thread needed no per-target
work — thread-local statics compile identically on all three
- admission is a pure env/host check ahead of toolchain discovery: an
iOS triple off a darwin host and near-miss mobile spellings refuse
SC3002 with the pairing or the supported set named, in compile() and
compileLibrary() both, decorated by the profile teaching machinery
- library-multi grows the mobile lanes: M12 admission shape always
runs; SCRIPTC_IOS=1 builds localized a/b archives per target and
emission (M1 symbol exactness, ambient audit, version floor), links
probes with the Xcode clang, and EXECUTES the two-instance and
four-thread probes on a booted simulator byte-for-byte against the
desktop expectations (device-arch archives are build+link verified);
SCRIPTC_ANDROID=1 does the same through the NDK clang and an
emulator, reusing a running device or writing and booting a headless
arm64 AVD directly
- the surface manifest's coverage notes record the mobile posture (the
library-admissible surface only; no mobile claim outside it), the
profile spec and the platforms page carry the mobile matrix with the
version floors, and non-mobile builds are unchanged byte-for-byte
* fix: harden mobile target tooling
* Harden mobile target test isolation
* feat: extend runtime symbol localization to Windows hosts and cross targets
- abi.localize_runtime follows the archive's OBJECT FORMAT instead of the
build host: ELF and COFF archives localize from any supported host
(native or cross), Mach-O localization keeps the macOS host linker, and
the SC3002 fence narrows to the pairings that remain — macos targets
off darwin hosts (named in the diagnostic) and hosts outside
darwin/linux/win32
- COFF combine+demote is an in-process transform (object-localize.ts):
no linker offers a COFF relocatable mode (lld-link mirrors MSVC
link.exe; zig's COFF driver refuses multi-object merges) and
llvm-objcopy rejects symbol-scope flags for COFF, so
mergeAndLocalizeCoffObjects performs the staging archive's member
selection, the section merge, cross-object symbol resolution by index,
COMDAT deduplication, and demotion to static over the object bytes;
survivors drop the COMDAT flag so mingw .refptr stubs never
deduplicate against another archive's copies at the embedder's link,
and per-object CodeView sections stay out of the merged member (their
item indices are a per-object contract)
- cross ELF merges through `zig cc -target <triple> -r` (zig is already
the cross driver's hard requirement) and localizeElfObject demotes in
process — locals-first symbol reorder, relocation remap, COMMONs kept
global — resolving section groups the way binutils'
--force-group-allocation does; the darwin ld64 and native-linux
binutils recipes are byte-for-byte the historical ones, and darwin
hosts now admit macos cross triples through the same ld64 arm
- library-multi grows the M10/M11 cross lanes (SCRIPTC_CROSS, with
container and Windows-box execution legs, thread-instanced composition
included), runs M1/M2/M6/M7 on win32 hosts (`zig cc` probes, llvm-nm
symbol checks, CRLF-normalized probe output), and object-localize.test.ts
pins the transforms over synthesized objects in every suite
- non-localized archives are unchanged byte-for-byte
* fix: tighten runtime object localization
* fix: honor COFF COMDAT selection semantics
* test: serialize cross-target library builds
* docs: clarify Windows library link dependencies
* feat: support thread-instanced library state
- Add the abi.instance_per_thread profile field: the archive's TUs compile
with -DSCR_THREAD_INSTANCES, and the SCR_TL qualifier in scr_runtime.h
moves every runtime unit's mutable statics into thread-local storage;
both backends emit the program TU's module globals, run-once guards, and
lazily-compiled regex literal caches thread-local to match
- ONE linked archive then serves one independent instance per embedder
thread through the unchanged entry family: a thread registers its sink
and calls the init entry, and owns its own collector, result arena,
poison flag, and program state — a trap poisons only the instance it
fired in while sibling threads' instances keep answering
- Document the contract beside the profile spec: one instance per thread,
selected implicitly by the calling thread; instance lifetime is the
thread's lifetime; the never-entered-from-two-threads rule is unchanged;
independent of and composable with abi.localize_runtime
- Immutable interned data (string literals, unit arms, template arrays,
vtables) stays shared — Darwin ASan's image-registration common
included, keeping the one-registration-per-image discipline; the
exception cell's current pointer resolves a NULL sentinel in this mode
because a thread-local address is not a constant initializer
- Add the four-thread acceptance probe (distinct per-thread workloads,
concurrent instance-local inits, per-instance collects, a trap delivered
to its own thread's sink exactly once), the composition probe pairing a
thread-instanced localized archive with a second different-prefix
localized archive, an explicit ASan rerun beside the suite-flavor
sanitized builds, and profile-shape coverage; schedule the new lane
contracts beside M1/M2
- Non-opted builds are byte-for-byte unchanged: the qualifier expands to
nothing outside -DSCR_LIB -DSCR_THREAD_INSTANCES (verified object-level
over every touched runtime TU in both default and library flavors)
* fix: keep sanitized runtime-localized archives linkable on ELF
ASan's instrumented globals ride ELF section groups; archives built from
shared runtime objects carry groups with REPEATED signatures, so a process
linking two runtime-localized sanitized archives kept one archive's group
and discarded the other's — whose now-local references then dangled at the
embedder's link. Resolving the groups into the combined relocatable member
(ld -r --force-group-allocation) keeps every archive's copies; the
localization step then demotes them per archive exactly like unsanitized
state. Plain builds carry no section groups, so the flag is inert there.
With the groups resolved, ELF surfaces the same deliberate exception
Mach-O already documented: the image-wide registration guard COMMON stays
shared so the final image registers its ASan globals exactly once — M1/M7
now pin that spelling on both platforms. The explicit ASan pairing (M8)
additionally points Linux LSan away from contractually thread-lifetime
instance state, exactly as the sanitized suite lanes do.
* fix: isolate inspect state per thread
* fix(runtime): share uptime anchor across threads
* feat: support multi-instance library mode via runtime symbol localization
- Add the abi.localize_runtime profile field: the archive build combines the
program object with exactly the runtime/vendor members it reaches and
demotes every external definition except the profile-declared symbols to a
local symbol (darwin: one ld -r pass with -exported_symbols_list; linux:
ld -r then objcopy --keep-global-symbols)
- N archives built under pairwise-distinct prefixes now link into one process
with no symbol collisions and no shared mutable runtime state: each
instance owns a private copy of the allocator, collector, result arena, and
panic sink, so sinks register per instance and a trap poisons only the
instance it fired in
- Document the embedder contract beside the profile spec: one thread per
instance (an instance is never entered from two threads), and values cross
instances only through the embedder's own byte/record marshalling
- Refuse cross-target localized builds with SC3002 (the step runs the host
toolchain's ld/objcopy over host-format objects); absent or false keeps the
classic single-archive artifact byte-for-byte
- Add the two-instance acceptance probe (two archives, two embedder threads,
independent collects, a trap delivered to its own sink exactly once while
the other instance keeps answering) plus symbol-exactness, profile-shape,
and target-posture suites, and schedule the artifact contracts on the gate
host
* fix: tighten runtime localization validation
* fix: harden localized library publication
Chris Tate