ci: macOS helper matrix + pinned shellcheck - #283
Merged
Merged
Conversation
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Advanced Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
acburdine
changed the title
acburdine
marked this pull request as ready for review
September 14, 2026 21:11
acburdine
force-pushed
the
claude/s2b-ci-macos
branch
from
September 14, 2026 21:13
4348744 to
6834a9f
Compare
…decode on bash 3.2 macOS is a supported local-dev host (`install.sh --local`), so the shell helpers run on its BSD userland and bash 3.2 — where a construct that passes on Linux or a newer bash can break. Nothing exercised that until now. - test.yml: add a macos-latest job running the suite under /bin/bash (3.2) with the Docker CLI but no daemon. Daemon-backed tests (mode matrix, Caddy, ingress, installer e2e) skip themselves via dockerAvailable(); the config-parsing and unreachable-daemon tests still run and are meaningful. Production ingress stays a Linux concern. - helpers.mjs: honor GD_TEST_BASH so the suite can be pointed at a specific interpreter; the macOS job sets it to /bin/bash. - shellcheck.yml: pin shellcheck to 0.10.0 (install the release) so CI and a developer's machine agree, instead of the runner's drifting apt version that disagreed with local on SC2015/SC2002. The new job immediately caught a real bash-3.2 bug it now guards against: - env.sh: the single-quote `\'` decode ran inside the printf's double quotes, where bash 3.2 and 4+ parse the pattern backslashes differently, so a single-quoted value read back wrong on macOS. Do the substitution in an unquoted assignment, which both agree on. Verified end to end under /bin/bash 3.2 (109 pass) and bash 5 (234 pass). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…s the docker CLI The macOS runner ships the Docker CLI (29.6.2) but no `docker compose` plugin, and `brew install docker-compose` does not register it as a `docker compose` subcommand, so the setup step failed on `docker compose version`. Install the plugin binary into ~/.docker/cli-plugins directly (pinned, darwin-aarch64), matching how the Linux job pins its Compose. No daemon involved — it is only used to parse configuration. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The prior "docker 29.6.2" came from the earlier `brew install docker`, not from the runner: macOS runners have no Docker at all, so dropping the install left `docker: command not found`. Restore `brew install docker` for the client (no daemon) and keep the direct compose-plugin download, since Homebrew's compose formula is not registered as a `docker compose` subcommand. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
acburdine
force-pushed
the
claude/s2b-ci-macos
branch
from
September 14, 2026 21:17
6834a9f to
b5350c6
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on S2 (#281). A focused CI/portability change kept out of S1/S2 so their diffs stay clean.
Why
The four portability bugs fixed across #280/#281 (BSD-vs-GNU
stat,docker infoexit code, shellcheck version skew, and now a bash-3.2 single-quote decode) were all invisible until CI ran on Linux — and one whole class (bash 3.2 + BSD userland) still had no coverage. macOS is a supported local-dev host (install.sh --local), so the shell helpers run there and must work.What
test.yml— newmacos-latestjob. Runs the suite under/bin/bash(3.2) with the Docker CLI installed but no daemon. Daemon-backed tests (mode matrix, Caddy, ingress, installer e2e) skip themselves viadockerAvailable(); the config-parsing and unreachable-daemon tests still run and are meaningful. Production ingress stays a Linux concern.helpers.mjs— honorsGD_TEST_BASH; the macOS job sets it to/bin/bashso a newer bash on the runner can't hide a 3.2 incompatibility.shellcheck.yml— pins shellcheck to 0.10.0 (installs the release) so CI and a developer's machine agree, instead of the runner's drifting apt version that disagreed with local 0.11 on SC2015/SC2002.It earned its keep on the first run
Caught a real bash-3.2 bug it now guards:
env.sh— the single-quote\'decode ran inside the printf's double quotes, where bash 3.2 and 4+ parse the pattern backslashes differently, so a single-quoted value read back wrong on macOS. Moved the substitution to an unquoted assignment, which both agree on.Verification
/bin/bash3.2 + no daemon (the macOS-runner state): 109 pass, 0 fail🤖 Generated with Claude Code