[pull] main from jdx:main - #140
Merged
Merged
Conversation
## Summary - When an install fails (e.g. GitHub rate limit 403, or a non-existent version 404), `cleanup_install_dirs_on_error` only removed the version-specific subdirectory (e.g. `installs/tilt/latest`) but left the parent directory (`installs/tilt/`) as an empty directory - Fix by also removing the parent `installs_path` when it is empty after cleanup, guarded so it won't remove it if other versions exist - Also cleans up the incomplete marker from cache (was also left behind on error) - Adds e2e regression test: installs a non-existent tilt version to trigger a 404 failure, asserts the parent dir is gone, then verifies a real install still works ## Test plan - [x] `mise run test:e2e test_aqua_failed_install_cleanup` passes - [x] Test fails without the patch (confirmed via `git stash`) 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- CURSOR_SUMMARY --> --- > [!NOTE] > **Medium Risk** > Touches install cleanup and config file selection logic; mistakes could delete unintended directories or write tool versions to the wrong config file, though changes are guarded and covered by new e2e tests. > > **Overview** > Fixes failed-install cleanup so `cleanup_install_dirs_on_error` also removes the cache “incomplete” marker and deletes the tool’s parent installs directory when it becomes empty, preventing leftover empty `installs/<tool>/` folders. > > Refines config write targeting by updating `first_config_file`/`global_config_path` to skip `conf.d` entries (and `.tool-versions`) so `mise use` and `mise use -g` prefer writing to `config.toml` over `config.local.toml`. Adds e2e coverage for both the failed-install directory cleanup regression and the global/local config write preference. > > <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit 901bf70. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup> <!-- /CURSOR_SUMMARY --> --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
## Summary PR #8279 placed every task child process in its own process group via `setpgid(0, 0)`. This broke interactive tools like Tilt that read from stdin. **Root cause:** The terminal driver only delivers keyboard input to the *foreground process group*. When a child is moved to a new process group, it's no longer the terminal's foreground group, so any attempt to `read()` from stdin causes the kernel to send `SIGTTIN`, which silently stops the process. Result: the process hangs, Ctrl-C has no effect, and the UI never loads. **Fix:** In `pre_exec`, check `std::io::stdin().is_terminal()`. If stdin is a TTY (inherited from the terminal), skip `setpgid` — the process stays in the terminal's foreground group and gets full keyboard I/O. If stdin is not a TTY (null/piped, i.e. non-interactive tasks), `setpgid` fires as before, preserving grandchild cleanup on Ctrl-C. This keeps the grandchild-kill benefit of #8279 for all normal (non-interactive) tasks while restoring correct behavior for interactive ones. Closes #8300 ## Test plan - [x] `cargo build` compiles cleanly - [x] Verified `mise run tilt:up` (or equivalent interactive task) no longer hangs - [x] Non-interactive tasks still get their own process group for clean teardown 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- CURSOR_SUMMARY --> --- > [!NOTE] > **Low Risk** > Small, Unix-only change gated on an `is_terminal` check; primary risk is behavior change in process-group/signal handling for commands with TTY stdin. > > **Overview** > Fixes Unix task execution for interactive commands by **only calling `setpgid(0,0)` when stdin is not a TTY**. > > In `CmdLineRunner::execute`’s `pre_exec` hook, stdin is checked via `BorrowedFd::borrow_raw(0)` + `IsTerminal`; interactive children stay in the terminal’s foreground process group to avoid `SIGTTIN` hangs, while non-interactive tasks keep the separate process group behavior for signal/cleanup handling. > > <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit 4bf5aef. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup> <!-- /CURSOR_SUMMARY --> --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
### 🐛 Bug Fixes - **(docs)** correct ripgrep command by @nguyenvulong in [#8299](#8299) - **(task)** skip setpgid for TTY stdin to fix interactive tasks by @jdx in [#8301](#8301) - clean up empty parent install dir on failed install by @jdx in [#8302](#8302) ### Chore - **(release)** run communique via mise x for PATH resolution by @jdx in [#8294](#8294) ## 📦 Aqua Registry Updates #### New Packages (2) - [`kubie-org/kubie`](https://github.com/kubie-org/kubie) - [`steipete/gogcli`](https://github.com/steipete/gogcli)
…itle (#8303) ## Summary - `communique generate` outputs the first line as a markdown heading (e.g. `# v2026.2.19: Fixes for...`) - `head -1` captured the `#` literally, making the GitHub release title `v2026.2.19: # v2026.2.19: Fixes for...` - Fix: pipe through `sed 's/^#* *//'` to strip the heading prefix before setting `RELEASE_TITLE` This caused the malformed title visible on https://github.com/jdx/mise/releases/tag/v2026.2.19. 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- CURSOR_SUMMARY --> --- > [!NOTE] > **Low Risk** > Single-line change to CI release-title formatting; minimal blast radius and no production code impact. > > **Overview** > Fixes GitHub release title generation in the `release` workflow by stripping leading Markdown heading characters from the first line of `communique generate` output before exporting `RELEASE_TITLE`. > > This prevents duplicated/malformed titles when `communique` emits a `# ...` heading on the first line. > > <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit bd7ca9e. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup> <!-- /CURSOR_SUMMARY --> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
## Summary - When `mise install --locked` is run, the `mise.lock` file should not be modified - Adds early returns in `update_lockfiles` and `auto_lock_new_versions` when `Settings::get().locked` is true - Consistent with how `cargo install --locked` and `uv pip install --locked` behave: the lock file is treated as authoritative and must not be updated ## Changes Made - `update_lockfiles()`: guard changed from `if !Settings::get().lockfile` to `if !Settings::get().lockfile || Settings::get().locked` - `auto_lock_new_versions()`: guard changed from `if !Settings::get().lockfile || new_versions.is_empty()` to `if !Settings::get().lockfile || Settings::get().locked || new_versions.is_empty()` ## Testing - Existing lockfile tests continue to pass - The `--locked` flag now prevents any writes to `mise.lock` during install ## Related - Spec: `/home/jdx/src/mise/src/lockfile.rs` <!-- CURSOR_SUMMARY --> --- > [!NOTE] > **Low Risk** > Small conditional-guard change that only disables lockfile writes in locked mode; minimal behavioral surface area and no security/data-handling impact. > > **Overview** > When `Settings::get().locked` (e.g. `mise install --locked`) is enabled, lockfile update paths now short-circuit to avoid modifying `mise.lock`. > > Specifically, `update_lockfiles` and `auto_lock_new_versions` add an early return when locked mode is active, preventing post-install lockfile refresh/auto-lock behavior. > > <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit fdccfd2. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup> <!-- /CURSOR_SUMMARY --> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Mapped `.vsix` extension to `TarFormat::Zip`. `vsix` archives are zip archives, without this mapping `.vsix`archives are treated as `raw` and therefore not extracted. tests : - `cargo test` - `Manual testing with actual `.vsix` file: ```toml [tools] 'http:java-debug'.version = "latest" 'http:java-debug'.url = "https://open-vsx.org/api/vscjava/vscode-java-pack/0.30.5/file/vscjava.vscode-java-pack-0.30.5.vsix" ```
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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
3 participants
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.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )