fix(build): make sa actually fail on shellcheck findings by Chemaclass · Pull Request #815 · TypedDevs/bashunit

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass · 2026-07-16T04:18:46Z

🤔 Background

make sa used find -exec shellcheck {} \;, and find swallows the command's exit code — the target printed findings yet always reported OK. Real findings (SC1087/SC2034 in #813) sailed through the local gate and only failed in CI.

💡 Changes

Run shellcheck one file per invocation via xargs -n 1 -P 4, which propagates failures (verified: a bad probe file now fails the target; clean tree passes in ~6s).
Mirror CI's SHELLCHECK_OPTS excludes so the local and CI gates check the same thing.
Per-file invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

'find -exec shellcheck {} \;' swallows shellcheck's exit code — find returns 0 regardless of findings, so 'make sa' printed warnings yet reported OK and let them through to CI (which then failed on SC1087/SC2034 that the local gate had 'passed'). Run one file per invocation via xargs, which propagates failures; -P 4 keeps it ~6s. The excludes mirror CI's SHELLCHECK_OPTS so both gates check the same thing. One file per invocation also sidesteps a shellcheck 0.11.0 crash (GHC pattern-match failure) on multi-file batches with -x.

Chemaclass added the bug Something isn't working label Jul 16, 2026

Chemaclass self-assigned this Jul 16, 2026

Chemaclass merged commit 6f43e6b into mainJul 16, 2026
37 checks passed

Chemaclass deleted the fix/make-sa-exit-code branch July 16, 2026 04:24