repro_issue_218.sh: cleanup can delete plugin folders another run made

Open
  • Is something wrong?
#590 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
56/100
Issue type
Bug
Clarity
Mostly clear
Activity status
Active
Tech stack
github-actions, lua, neovim, shell
Domain
devtools, testing

Research direction

Start with scripts/repro_issue_218.sh and trace how bootstrap cloning and signal cleanup handle entries in the shared lazy folder. Add a network-free stubbed regression script covering both reproductions, then wire it into CI. The issue describes atomic bootstrap cloning as the planned fix; done means the new regression fails on the current behavior and passes with the fix, including concurrent runs and signal cleanup.

Written by the indexing model from the issue text.

Description

needs-triage

scripts/repro_issue_218.sh cleanup can delete plugin folders that another run made (follow-up to a Codex P2 on #458).

Defect. On SIGTERM, SIGINT or SIGHUP during Lazy! restore, the cleanup removes every entry in <data>/issue-218/lazy that was not there before the restore. A plugin folder that another run (or vv issue-218) created in the same data folder in the meantime is deleted too. Only public plugin clones are affected, not user data, and it needs a concurrent cold install plus a signal.

Reproductions.

  • A: start the script, then during the restore add lazy/other.nvim (with a .cloning marker) and a full snacks.nvim copy to the same lazy folder, then signal the script or its process group. Main deleted both folders in 6 of 6 trials.
  • B: two runs on one data folder, run A holds lazy.nvim's bootstrap clone. On main, run B exits 2 (a half clone sits under the real name lazy), and run A's exit leaves lazy.nvim invalid.

Planned fix. Atomic bootstrap: clone into lazy.nvim.bootstrap-<run> and rename it when the clone is complete, so cleanup removes only that one name. In a trial it passed A (6 of 6 folders survive) and B, and the cleanup got shorter. Add a stubbed regression script (no network) that fails on main and passes with the fix, and wire it into CI. The same branch can also refuse a TMPDIR with $ ' \ , { [`, use job control so a backgrounded run still stops its Neovim, and document the side effects.

Owner: the next change to scripts/repro_issue_218.sh. Deferred from the README verification run with the maintainer's agreement.

Generated with xum • Model: anthropic:claude-opus-5-5 • Thinking: high

Dominant language
Lua
Stars
3.2k
Forks
223
Avg merge
6h 7m
Merged PRs (30d)
210

Getting set up

Open in Codespaces

Starts the project's dev container in your browser, under your own GitHub account.

  • No Dockerfile or Docker Compose file
  • No pull request template
  • No contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from coder/claudecode.nvim

All issues in coder/claudecode.nvim

Similar issues

More Lua issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.