Tom Boucher 9263fa1e46 test(#606): regression guard — every same-dir require() target of a shipped hook must itself ship (#613) ...

* test(#606): guard that every same-dir require() target of a shipped hook is shipped

#606: gsd-check-update-worker.js require()'d its sibling managed-hooks-registry.cjs,
but the file was missing from HOOKS_TO_COPY, so the installer never placed it next
to the worker and the background update worker crashed silently with
"Cannot find module". The fix shipped in #611 (added the file to HOOKS_TO_COPY);
this adds the regression guard that was the issue's third acceptance criterion.

The new test scans every JS/CJS hook in HOOKS_TO_COPY for same-directory relative
requires — require('./x') and require('./subdir/x') — and asserts each target is
itself shipped: './x' is in HOOKS_TO_COPY, or './subdir/...' lives under a dir in
HOOKS_SUBDIRS_TO_COPY. require('../...') targets that escape the hooks/ dir are out
of scope (their shipping is governed by package.json "files").

Exports HOOKS_SUBDIRS_TO_COPY from scripts/build-hooks.js so the test reads the real
subdir allowlist instead of hardcoding ['lib'].

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* chore(#606): add changeset crediting installer registry fix

Fixed-type fragment so the #606 installer fix (managed-hooks-registry.cjs
now shipped) is credited in the release notes. The behavioral change landed
in #611; this records it for the changelog and credits the reporter.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-02 13:00:29 -04:00

9.9 KiB

Raw Blame History

View Raw