perl: run on the default/unset locale without a startup panic (kd-dvph) by brandonpayton · Pull Request #822 · Automattic/kandelo

brandonpayton · 2026-07-01T12:42:15Z

Purpose

Perl on Kandelo aborted at interpreter startup for any locale except literally LC_ALL=C, so perl -e 'print 2+3' panicked (exit 29) with the default/unset locale or any UTF-8 locale. This makes Perl usable on Kandelo with a normal environment. Fixes kd-dvph (blocks the homebrew-all umbrella kd-1mr).

Root cause

Kandelo's libc is musl, whose setlocale(LC_ALL, "") returns a positional, ;-separated composite of the per-category locales in category order (CTYPE;NUMERIC;TIME;COLLATE;MONETARY;MESSAGES) when categories differ — e.g. with no locale env it returns C.UTF-8;C;C;C;C;C. This is POSIX-legal (the LC_ALL return string is unspecified and need only round-trip through setlocale).

perl-cross cross-configured the target with glibc's assumption (d_perl_lc_all_uses_name_value_pairs=define), so perl compiled out its positional parser and parsed musl's first field C.UTF-8 as a glibc name=value pair. No = → locale.c panic "needs an '=' to split name=value" → exit 29.

This is a perl↔musl compatibility-boundary issue in the port config, not a platform gap: musl's format is POSIX-conformant, and forcing musl to emit glibc's format would change libc-wide behavior for every setlocale caller.

Fix

build-perl.sh patches the target config.h (the primary perl.wasm config; the build-time miniperl keeps xconfig.h) to perl's positional LC_ALL mode — the mechanism perl already ships for *BSD:

#undef PERL_LC_ALL_USES_NAME_VALUE_PAIRS
#define PERL_LC_ALL_SEPARATOR ";"
#define PERL_LC_ALL_CATEGORY_POSITIONS_INIT { LC_CTYPE, LC_NUMERIC, LC_TIME, LC_COLLATE, LC_MONETARY, LC_MESSAGES } (musl's emit order)

get_category_index() maps each position to perl's internal index; a compile-time STATIC_ASSERT guards the count. build.toml revision 1 → 2 (output bytes change).

Verification (both kernel hosts)

Node (runCentralizedProgram + NodePlatformIO) — 8 pass / 0 fail / 1 skip; browser (headless Chromium via BrowserKernel) — 2 pass / 0 fail. perl -e now runs with an unset locale, LC_ALL=C, LC_ALL=C.UTF-8, LANG=C.UTF-8, LANG=en_US.UTF-8 (the browser default), and a disparate composite — all exit-29 panics before. Built-in ${^UTF8LOCALE} confirms per-category parsing routes CTYPE correctly (not just non-panicking).

New regression smokes: packages/registry/perl/demo/locale-smoke.ts (Node) and locale-browser-smoke.ts (browser). Docs: a porting-guide troubleshooting note on the musl-positional vs glibc-name=value LC_ALL format.

Package-recipe-only change (build script + demo + docs); no kernel/host/musl/ABI code, so the 5-suite full gate is not triggered.

Coordination

Overlaps build-perl.sh with the active perl-runtime work in #821 (kd-k7zy); this patch is orthogonal (startup locale scan vs module-loading) and small. Merge sequencing to be coordinated.

🤖 Generated with Claude Code

The Perl bottle shipped only perl.wasm with no core-module runtime, and the interpreter crashed on any real module load, so the port only ever passed a trivial arithmetic smoke. Make Perl-on-Kandelo usable: File::Spec, POSIX, Cwd, XSLoader, Config and the pure-perl core all load and run. Fixes kd-k7zy (the kd-p3hr follow-up where File::Spec failed on missing generated XSLoader.pm). Three root causes, all in the build recipe: - Generated runtime files were never produced/packaged: `make perl` stops before perl-cross's nonxs/extensions targets that generate lib/XSLoader.pm and stage the core tree, and the target Config had an empty osname so MakeMaker's probes failed. Fix: -Dosname=linux + build with `make -k`, then package lib/perl5/5.40.3 as a new perl-runtime.zip output. - The interpreter miscompiled and panicked on any weak-ref use: -O2 without -fno-strict-aliasing miscompiles perl's weak-ref code into a magic_killbackrefs panic. The flag was on HOSTCFLAGS only; add it to the target -Dccflags. - XS core modules could not load: usedl=define builds .so's loaded via dlopen, which Kandelo wasm lacks. -Uusedl links XS into perl.wasm with a boot table so XSLoader::load resolves them without dlopen. perl-cross ignores -Dnoextensions, so the static set is curated by editing Makefile.config's fullpath_static_ext (drop ext/re, which duplicates core regcomp symbols, and external-lib/threads exts). Also work around a perl-cross bug where the static-ext recipe never runs pm_to_blib: after make -k, stage each curated static ext's .pm. build.toml revision 1 -> 2 (output bytes change). package.toml declares the perl-runtime output. Formula/perl.rb installs perl.wasm + the runtime and tests File::Spec/POSIX/XSLoader. demo/runtime-smoke.ts is a Node kernel-host smoke. Verified via Node runCentralizedProgram against the packaged perl-runtime.zip (LC_ALL=C): File::Spec catfile/rel2abs, Cwd, POSIX, Fcntl, List::Util, Data::Dumper, XSLoader, Config all load (PERL_RUNTIME_SMOKE_PASS). Known boundaries filed: kd-dvph (default-locale startup panic), kd-gtxa (Errno errno extraction), kd-14n8 (broaden static XS / dlopen). Browser bottle tracks kd-yuef. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

github-actions · 2026-07-01T14:44:35Z

Phase B-1 matrix build status — `pr-822-staging`

ABI v16. 67 built, 0 failed, 67 total.

Package	Arch	Status	Sha
libcurl	wasm32	built	`b01be826`
libcxx	wasm32	built	`bd2dd5a6`
libcxx	wasm64	built	`926cfc96`
libpng	wasm32	built	`25e56ef0`
libxml2	wasm32	built	`eb77d16a`
libxml2	wasm64	built	`9743c3c9`
openssl	wasm32	built	`296793ae`
openssl	wasm64	built	`b53a16d9`
sqlite	wasm32	built	`93b98c80`
sqlite	wasm64	built	`7079a1ac`
zlib	wasm32	built	`4543740c`
zlib	wasm64	built	`4ccf5221`
bc	wasm32	built	`0c54e287`
bzip2	wasm32	built	`614536d6`
coreutils	wasm32	built	`e1a33298`
curl	wasm32	built	`ecbc3967`
dash	wasm32	built	`4f2caf8b`
diffutils	wasm32	built	`34f174f8`
dinit	wasm32	built	`97a7849b`
fbdoom	wasm32	built	`3799bb5e`
file	wasm32	built	`535dd689`
findutils	wasm32	built	`4b727dbf`
gawk	wasm32	built	`92e5184d`
git	wasm32	built	`42b4d1cb`
grep	wasm32	built	`fd4d79fa`
gzip	wasm32	built	`8f54d8ac`
kandelo-sdk	wasm32	built	`ba15af22`
kernel	wasm32	built	`917959d2`
less	wasm32	built	`7f3184f8`
lsof	wasm32	built	`4dc5ae5b`
m4	wasm32	built	`5fbbe8b3`
make	wasm32	built	`57a6d854`
mariadb	wasm32	built	`9ed4aa94`
mariadb	wasm64	built	`eaf29411`
modeset	wasm32	built	`d9f12284`
msmtpd	wasm32	built	`23172a3a`
nano	wasm32	built	`763f56f8`
ncurses	wasm32	built	`a7862ee0`
netcat	wasm32	built	`57bdd0cd`
nginx	wasm32	built	`933bdcd3`
php	wasm32	built	`e047ea3c`
posix-utils-lite	wasm32	built	`2aec933c`
sed	wasm32	built	`9d958a03`
spidermonkey	wasm32	built	`ef0ac7d5`
tar	wasm32	built	`050e7cb3`
tcl	wasm32	built	`ee9ae67a`
unzip	wasm32	built	`b176c19f`
userspace	wasm32	built	`a9e013a2`
vim	wasm32	built	`8194ff88`
wget	wasm32	built	`378dca44`
xz	wasm32	built	`763b854c`
zip	wasm32	built	`1c834314`
zstd	wasm32	built	`bcd693e5`
bash	wasm32	built	`d435c352`
mariadb-test	wasm32	built	`74e71062`
mariadb-vfs	wasm32	built	`85427981`
mariadb-vfs	wasm64	built	`6c3ac528`
nethack	wasm32	built	`5b7fc658`
node	wasm32	built	`cdad27ac`
spidermonkey-node	wasm32	built	`6d8dc738`
vim-browser-bundle	wasm32	built	`ef862c6f`
nethack-browser-bundle	wasm32	built	`d4992e6b`
rootfs	wasm32	built	`ae3df667`
shell	wasm32	built	`61c3bb81`
lamp	wasm32	built	`f5445d80`
node-vfs	wasm32	built	`75f6e69a`
wordpress	wasm32	built	`1c1bcff5`

_{Auto-generated; replaced on each push. Raw data in the publish-status workflow artifact.}

ext/Errno/Errno_pm.PL discovers the E* errno constants by preprocessing `#include <errno.h>` and scanning cpp output for `# <line> "file"` linemarkers. perl-cross defines cpp/cpprun/cppstdin as "$cc -E -P"; -P suppresses those linemarkers, so on the wasm cross target the scan found no headers, collected no constants, and Errno_pm.PL died "No error definitions found" -- Errno.pm was never generated/staged and `use Errno` failed. The constants are plain `#define E* <int>` in the sysroot (musl arch/generic bits/errno.h). build-perl.sh now patches get_files() to fall back to the sysroot errno headers when linemarker discovery yields nothing, adds Errno.pm to the fail-loud runtime post-check, and bumps the package revision (perl-runtime.zip now ships Errno.pm; perl.wasm is byte-unchanged). Adds Node + browser use-Errno smokes. Stacked on #821 (kd-k7zy). Verified: Node errno-smoke 16/16 (Errno.pm with 134 musl-valued constants + %! tie); browser smoke harness committed (skips where the browser bundle is absent; browser acceptance tracked by kd-yuef). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

musl's setlocale(LC_ALL,"") returns a positional ';'-separated composite (e.g. "C.UTF-8;C;C;C;C;C") when the categories differ, but perl-cross configured the wasm target with glibc's name=value assumption (PERL_LC_ALL_USES_NAME_VALUE_PAIRS). perl parsed musl's first field "C.UTF-8" as name=value, found no '=', and aborted at interpreter startup (exit 29) for every locale except literally LC_ALL=C -- making perl unusable with the default or any UTF-8 locale on Kandelo. build-perl.sh now patches the target config.h to perl's positional LC_ALL mode with musl's category order (CTYPE;NUMERIC;TIME;COLLATE;MONETARY;MESSAGES), the mechanism perl already ships for *BSD; get_category_index() maps each position to its internal index and a STATIC_ASSERT guards the count. Host miniperl (xconfig.h) is left matching the build machine's libc. build.toml revision 1->2 (output bytes change). Adds a Node + browser locale smoke and a porting-guide troubleshooting note. Verified on both kernel hosts: perl -e runs with an unset locale, LC_ALL=C, LC_ALL=C.UTF-8, LANG=C.UTF-8, LANG=en_US.UTF-8, and a disparate composite (all exit-29 panics before); built-in ${^UTF8LOCALE} confirms per-category parsing routes CTYPE correctly. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

brandonpayton · 2026-07-02T02:46:31Z

Re-stacked as part of the perl convoy. Base is now gascity/kd-1mr/kd-gtxa-errno-constants (#827, Errno), which is itself stacked on #821 (kd-k7zy runtime). Linear stack: #821 → #827 → #822.

This PR's own change is the 5 locale files only: build-perl.sh (target config.h musl positional LC_ALL patch), build.toml (revision → 4), docs/porting-guide.md (locale gotcha), and the two locale smoke demos. Per GitHub's compare API the range #827...#822 is 1 commit / 5 files (ahead 1, behind 0).

Note: the 'Files changed' tab may briefly show the whole stack (~10 files) — that's a base-change cache lag (the head was pushed just before the base was retargeted); it recomputes on its own and will retarget to main when #827 merges. Merge order: #821, then #827, then this PR.

brandonpayton force-pushed the gascity/kd-1mr/kd-dvph-perl-default-locale-startup-panic-composite-lc-all-unpar branch from f0bf4f8 to 92022d4 Compare July 1, 2026 12:44

brandonpayton mentioned this pull request Jul 1, 2026

perl: generate Errno.pm on the wasm cross target so use Errno works (kd-gtxa) #827

Open

brandonpayton changed the base branch from main to gascity/kd-1mr/kd-gtxa-errno-constants July 2, 2026 02:42

brandonpayton force-pushed the gascity/kd-1mr/kd-dvph-perl-default-locale-startup-panic-composite-lc-all-unpar branch from 92022d4 to 392e9c0 Compare July 2, 2026 02:42

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

perl: run on the default/unset locale without a startup panic (kd-dvph)#822

perl: run on the default/unset locale without a startup panic (kd-dvph)#822
brandonpayton wants to merge 3 commits into
gascity/kd-1mr/kd-gtxa-errno-constantsfrom
gascity/kd-1mr/kd-dvph-perl-default-locale-startup-panic-composite-lc-all-unpar

brandonpayton commented Jul 1, 2026

Uh oh!

github-actions Bot commented Jul 1, 2026

Uh oh!

brandonpayton commented Jul 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

brandonpayton commented Jul 1, 2026

Purpose

Root cause

Fix

Verification (both kernel hosts)

Coordination

Uh oh!

github-actions Bot commented Jul 1, 2026

Phase B-1 matrix build status — pr-822-staging

Uh oh!

brandonpayton commented Jul 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Phase B-1 matrix build status — `pr-822-staging`