fix(doctor): §25 crashed under macOS system bash 3.2 — backtick in $()-nested quoted heredoc (BRO-1718)#93
Conversation
…)-nested quoted heredoc (BRO-1718) v0.34.0's §25 script_path() heredoc carried a literal backtick in the operator-blacklist tuple (+4 in comments). The delimiter is quoted, but the block lives inside $( ... ) and bash 3.2's pre-parser dies on backticks there — and, as the fix round surfaced, on unbalanced quotes and dollar-paren openers in raw body text too. Shebang is #!/bin/bash → stock macOS = 3.2 → doctor died at §25 with no verdict for exactly the fresh-install users BRO-1715 targeted. Four green verification layers (dev wrapper, ubuntu CI, local suite, hermetic fresh-install test) all resolve bash from a modern PATH — one shared premise, bash >= 4; the delivery interface (direct shebang exec) was the one path nobody exercised. Infrastructure-layer instance of shared-assumption-drift. - doctor.sh: all 5 backticks purged from the §25 body (tuple → chr(96), comments → plain quotes, NB documents the constraint). Zero bash-4-isms → fully 3.2-clean: direct shebang exec now runs all 25 sections to a verdict. - tests/bash32-parse-safety.test.sh: class guard, RED on 0.34.0 → GREEN. A1: no backtick in $()/<( -nested quoted-heredoc bodies (scripts/tests/bin; scanner builds parser-hostile chars via chr() — it scans itself). A2: /bin/bash -n parses everything — real 3.2 on macOS dev machines. P20 cross-review: 6/10 round 1 (the guard itself carried the bug it polices — caught adversarially), fix round applied; verified vs validate-release.yml. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Warning Review limit reached
Next review available in: 13 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (4)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
1 participant
What
v0.34.1 — patch release.
doctor.sh§25 (shipped hours ago in v0.34.0) dies at parse time under stock macOS/bin/bash3.2 (the script's own shebang):The §25
script_path()python heredoc carried a literal backtick in the operator-blacklist tuple (+4 more in comments). The delimiter is quoted (<<'PYEOF') but the block sits inside$( ... ), and bash 3.2's command-substitution pre-parser scans the raw body text and dies on backticks there — and (surfaced during the fix round) on unbalanced quotes and dollar-paren openers too.Why every gate missed it
Dev wrapper (
exec bash→ PATH bash 5.3), ubuntu CI (bash 5), local suite (bash "$t"→ PATH bash), hermetic fresh-install test (bash "$REPO/scripts/doctor.sh") — four green layers, one shared premise: bash ≥ 4. The delivery interface (direct shebang execution — exactly how a stock-macOS fresh-install user runs it) was the one path nobody exercised. Infrastructure-layer instance ofpattern/shared-assumption-drift, sibling of 0.34.0'spattern/self-hosting-vacuous-pass.Fix
chr(96)(behaviorally identical membership test), comment backticks → plain quotes, NB comment documents the constraint. Zero bash-4-isms in the file → fully 3.2-clean. Verified: direct shebang execution under 3.2.57 runs all 25 sections to a verdict against the self-hosting workspace; §-by-§ output identical to bash 5.3.$(/<(, scanningscripts/+tests/+bin/. Standalone heredocs deliberately unflagged (parse-safe in 3.2; compute-arc-status.sh carries them today). Documented residual: multi-line$(spans — backstopped by A2 on macOS./bin/bash -nparses every script — under the system bash: a real 3.2 on macOS dev machines, i.e. the local suite now exercises the delivery interface ubuntu CI cannot.P20 cross-review (Strata B, adversarial)
Round 1: 6/10 — reviewer proved the first version of the guard itself carried a literal backtick inside its own
$()-nested heredoc (the scan's comparison line): the guard embodied the bug it polices and couldn't parse on the machines it protects. Fix round applied: scanner builds every parser-hostile char viachr(),tests/added to its own scan scope, scanner exit-status checked (L1),bin/shebang-filtered (L2),<(+double-quoted-delimiter forms covered and residuals documented honestly (M1). Negative control re-proven after the round.Validation
validate-release.ymlgates: VERSION 0.34.0→0.34.1 semver-monotonic, CHANGELOG section matchesLinear: BRO-1718
🤖 Generated with Claude Code