Repository navigation
fix(task): normalize task cwd for source freshness - #11987
Merged
Merged
Conversation
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository YAML (base), Central YAML (inherited), Organization UI (inherited) Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
📝 WalkthroughWalkthroughThe change normalizes task-specific working directories by removing ChangesTask working-directory freshness
Estimated code review effort: 2 (Simple) | ~10 minutes Mergeability Score: ⚪ Minimal · up to This localized change normalizes task working-directory paths so valid source files participate in freshness checks, with regression coverage included; no actionable merge-blocking risk remains beyond normal checks and review. Possibly related PRs
Suggested reviewers: Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
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 |
Contributor
Greptile SummaryNormalizes lexical current-directory components in task working directories so relative source paths and glob results share the same representation during freshness checks.
Confidence Score: 5/5The PR appears safe to merge; no concrete correctness or security regressions were identified. Removing lexical current-directory components preserves resolved path semantics while aligning the task cwd with normalized source paths, and the added unit and end-to-end tests cover both rerun and fresh-skip behavior. Important Files Changed
Reviews (1): Last reviewed commit: "fix(task): normalize task cwd for source..." | Re-trigger Greptile |
jdx
enabled auto-merge (squash)
August 14, 2026 01:47
stevenpollack
added a commit
to stevenpollack/mise
that referenced
this pull request
Aug 14, 2026
Revert the task-cwd half of the normalizer reconciliation. `task_cwd`'s result is passed to `Command::current_dir`, so resolving `..` there lexically changes where the task actually runs: with `dir = "link/.."` mise executed in `<ws>/real` before and `<ws>` after, because the OS resolves `..` through the symlink at chdir time and a lexical pass does not. Collapsing `.` is safe because `a/./b` and `a/b` are the same directory to both. The `task_state_key` de-duplication that motivated the change was a convenience riding along on a bug fix and is not worth a silent change to where commands execute; nothing in the source-anchoring fix depends on it, since `normalize_pattern` normalizes the joined path itself. `normalize_task_cwd` is left exactly as jdx#11987 wrote it, and `lexical_normalize` now documents that it is for paths that are matched, not for a directory that gets chdir'd into. Inspect the pattern as written in the glob-collapse guard. Checking the path already joined onto `task_cwd` folded the task directory into the string being scanned, so a task in a directory whose literal name contains a glob metacharacter — `dir = "pkg*"` — made an ordinary `../shared/**` entry look like a `..` popping a glob. The pattern was then left unanchored and matched nothing. The guard only ever needed to see what the user wrote. Correct the `sources` documentation, which said relative entries resolve against the config file. They resolve against the task directory, which is the config file's directory only when the task has no `dir`. The separate rule for `[task_config.input_groups]` entries, which do resolve against their defining config, is documented where those are described and is unaffected. Co-Authored-By: Claude <noreply@anthropic.com>
stevenpollack
added a commit
to stevenpollack/mise
that referenced
this pull request
Aug 14, 2026
Revert the task-cwd half of the normalizer reconciliation. `task_cwd`'s result is passed to `Command::current_dir`, so resolving `..` there lexically changes where the task actually runs: with `dir = "link/.."` mise executed in `<ws>/real` before and `<ws>` after, because the OS resolves `..` through the symlink at chdir time and a lexical pass does not. Collapsing `.` is safe because `a/./b` and `a/b` are the same directory to both. The `task_state_key` de-duplication that motivated the change was a convenience riding along on a bug fix and is not worth a silent change to where commands execute; nothing in the source-anchoring fix depends on it, since `normalize_pattern` normalizes the joined path itself. `normalize_task_cwd` is left exactly as jdx#11987 wrote it, and `lexical_normalize` now documents that it is for paths that are matched, not for a directory that gets chdir'd into. Inspect the pattern as written in the glob-collapse guard. Checking the path already joined onto `task_cwd` folded the task directory into the string being scanned, so a task in a directory whose literal name contains a glob metacharacter — `dir = "pkg*"` — made an ordinary `../shared/**` entry look like a `..` popping a glob. The pattern was then left unanchored and matched nothing. The guard only ever needed to see what the user wrote. Correct the `sources` documentation, which said relative entries resolve against the config file. They resolve against the task directory, which is the config file's directory only when the task has no `dir`. The separate rule for `[task_config.input_groups]` entries, which do resolve against their defining config, is documented where those are described and is unaffected. Co-Authored-By: Claude <noreply@anthropic.com>
stevenpollack
added a commit
to stevenpollack/mise
that referenced
this pull request
Aug 16, 2026
Revert the task-cwd half of the normalizer reconciliation. `task_cwd`'s result is passed to `Command::current_dir`, so resolving `..` there lexically changes where the task actually runs: with `dir = "link/.."` mise executed in `<ws>/real` before and `<ws>` after, because the OS resolves `..` through the symlink at chdir time and a lexical pass does not. Collapsing `.` is safe because `a/./b` and `a/b` are the same directory to both. The `task_state_key` de-duplication that motivated the change was a convenience riding along on a bug fix and is not worth a silent change to where commands execute; nothing in the source-anchoring fix depends on it, since `normalize_pattern` normalizes the joined path itself. `normalize_task_cwd` is left exactly as jdx#11987 wrote it, and `lexical_normalize` now documents that it is for paths that are matched, not for a directory that gets chdir'd into. Inspect the pattern as written in the glob-collapse guard. Checking the path already joined onto `task_cwd` folded the task directory into the string being scanned, so a task in a directory whose literal name contains a glob metacharacter — `dir = "pkg*"` — made an ordinary `../shared/**` entry look like a `..` popping a glob. The pattern was then left unanchored and matched nothing. The guard only ever needed to see what the user wrote. Correct the `sources` documentation, which said relative entries resolve against the config file. They resolve against the task directory, which is the config file's directory only when the task has no `dir`. The separate rule for `[task_config.input_groups]` entries, which do resolve against their defining config, is documented where those are described and is unaffected. Co-Authored-By: Claude <noreply@anthropic.com>
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
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.
Summary
dir = "./relative"Root cause
The task working directory retained the lexical
./component, while glob expansion returned normalized source paths. The source matcher therefore filtered out valid task-directory-relative sources, leaving only the task definition in the freshness hash and causing changed inputs to be skipped.Fixes the bug reported in #11977.
Validation
cargo test --bin mise task::task_source_checker::tests::relative_sources_match_when_task_dir_starts_with_dot -- --exactmise run test:e2e e2e/tasks/test_task_source_freshness_with_cwdmise run lint-fixcargo clippy --workspace --all-features --all-targets -- -D warningsAI-assisted — Tool: Codex; model: openai/gpt-5; version: unavailable.
Note
Low Risk
Small, localized path normalization in task CWD resolution with targeted tests; no auth, data, or broad API changes.
Overview
Fixes source/output freshness when a task uses a static
dirwith a leading./(e.g.dir = "./relative"). The effective working directory kept a lexical.segment while enumerated source paths were normalized, so the source matcher dropped task-relative files and tasks could stay skipped after inputs changed.task_cwdnow strips current-directory (./) components vianormalize_task_cwdwhile leaving..and other path semantics unchanged. Regression coverage adds a unit test for./suband an e2e scenario forrelative_dirwithsources/outputs.Reviewed by Cursor Bugbot for commit e187af7. Bugbot is set up for automated code reviews on this repo. Configure here.
Summary by CodeRabbit
Bug Fixes
./.Tests