[Question] Why not use git stash save --keep-index

[drmercer]

Thanks for the great tool! I've noticed something that might be a way to improve it. Whenever I have partially-staged files, the "Stashing changes..." part takes a frustratingly long time, as does the "Restoring local changes..." part afterwards. It also makes git status have to take several seconds to refresh the index first when I use it after lint-staged runs.

I looked into the source and found the lines that do the stashing:
https://github.com/okonet/lint-staged/blob/63b085f7b8a6ae953c0070fcb4122499fdb73542/src/gitWorkflow.js#L67-L82

Is there a reason these can't be replaced by a call to
git stash save --keep-index 'Automatic stash by lint-staged', and then git stash pop to unstash?

(I'd be happy to open a PR if that's a viable change.)

[drmercer]

I found #583 which talks about this more. Still unsure why the current solution works better than git stash --include-untracked --keep-index (which could introduce conflicts apparently). How does the current solution avoid introducing conflicts? 🤔 I'll do some experimenting if I have time.

[okonet]

Hey @drmercer sorry for a late reply.

Of course git stash -k was the first thing I tried a few years ago but the problem with internal stash is that if there are conflicts then it's going to fail the stash pop operation with a very cryptic message. In order to avoid this I'm using a low level approach proposed by someone in the discussion to the issue.

The way it works now is described in the PR. The important bit is:

Compare new commit index and working copy and apply as many formatting changes to working copy as possible. If there are conflicting hunks, drop formatter's changes. User modifications that were stashed should always take precedence over formatters.

And this is where it's implemented: https://github.com/okonet/lint-staged/pull/75/files#diff-cf453db1a0c4f207753345e32ef28aa3R121

Feel free to experiment since there is a robust test coverage (I've invested a lot of time into TDD for this one). I'd be open for a better or more "cleaner" solution to the problem but I could not find one.

[okonet]

🎉 This issue has been resolved in version 9.5.0-beta.1 🎉

The release is available on:

• npm package (@beta dist-tag)
• GitHub release

Your semantic-release bot 📦🚀

[okonet]

🎉 This issue has been resolved in version 10.0.0 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀