Repository navigation
fix(squash): back up the index with plumbing so hidden submodule changes don't abort the squash - #4303
Merged
Conversation
…ges don't abort the squash Since has_staged_changes reads the index with plumbing (#4192), a staged submodule pointer bump hidden by `submodule.<name>.ignore=all` counts as staged, so the squash takes its backup path. The backup ran `git stash create`, which honors that config, saw nothing, and aborted the squash with "git stash create returned empty SHA". The backup now writes the index with `write-tree` and `commit-tree -p HEAD`, so it holds exactly what the squash commits on top of the pre-squash tip, whatever the porcelain config hides. Unstaged and untracked files were never at risk: the squash leaves them in place. The backup commit is unsigned. Backup SHAs in nine snapshots change with the new commit shape. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
worktrunk-bot
approved these changes
Sep 27, 2026
Merged
max-sixty
added a commit
that referenced
this pull request
Sep 27, 2026
Bumps 0.79.0 → 0.80.0 and adds the `## 0.80.0` changelog section, cut from `12efb5159`. The bump is minor because the library API breaks (see Verification). ## Verification - `cargo run -- hook pre-merge --yes` on the cut tip: 4985 passed. - `nightly.yaml` on `12efb5159`, all jobs green including feature-powerset: https://github.com/max-sixty/worktrunk/actions/runs/36338233859 - `cargo semver-checks check-release -p worktrunk` reports five major-level lints, the removals and signature changes listed in the Internal entry. - Changelog: repeated verification passes against source until one came back clean. Fixed headlines name the condition a bug needed rather than its worst case, and near-zero-reach fixes are folded into two roll-up bullets, per #4295. - Data-loss surface review over `v0.79.0..HEAD` with independent finders. The one real finding, a nested stale detached worktree read as the enclosing repo's rebase so `wt step prune` could unregister it, was fixed in #4302. The review also found a squash regression in the backup path, fixed in #4303. The remaining candidates were adjudicated acceptable and signed off: #4228 unregistering stale worktrees whose directory remains (files stay on disk; the branch is deleted only if integrated), #4206 leaving a detached `HEAD` on Ctrl-C during a slow commit hook, #4251 allowing `--no-squash` merges with untracked files, #4219 dropping `--no-hooks` from the skill example, and #4279's `docker stop` recipe. #4252 closes a submodule data-loss path. > _This was written by Claude Code on behalf of max-sixty_ 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
1 task
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.
#4192 (unreleased) made
has_staged_changessee a staged submodule pointer hidden bysubmodule.<name>.ignore=all, which sends such a squash into the safety backup. The backup usedgit stash create, which honors that ignore setting, saw nothing, and returned an empty SHA, sowt step squashand a default-squashwt mergeaborted with✗ git stash create returned empty SHA - no changes to backup. In 0.79.0 a multi-commit squash in that setup succeeded (leaving the pointer out), so this would ship as a regression. The 0.80.0 changelog verification found it.The backup now snapshots the index with plumbing —
write-tree, thencommit-tree --no-gpg-sign <tree> -p HEAD, recorded inrefs/wt-backup/<branch>'s reflog as before — so no config hides part of it. By the time the backup runs, squash has already staged per its stage mode, so the index is exactly what the squash absorbs; unstaged and untracked files, which the squash leaves in place, are no longer in the backup.--no-gpg-signkeeps the backup from prompting for a signature, asstash createnever signed.test_step_squash_backs_up_submodule_bump_hidden_by_ignorecovers the two-commit, one-commit, and commits-cancel-out shapes, each with a hidden staged pointer: all three failed with the stash error before, and now squash including the pointer, with the backup's tree equal to the squash tree and its parent the pre-squash tip. Nine snapshots change only on theirBackup created @ <sha>line.