Stash changes not in index before applying hooks

[Northo]

Hi, and thanks for developing this great tool!

Unexpected behavior: When applying the hooks with Mookme, the hooks are run with the working tree as-is, as far as I understand. This may cause confusing behavior for files that are only partially stashed. For example, a linter may pass even when the stashed file is not acceptable.

What do you suggest: pre-commit solves this by stashing away the working tree not in index before the hooks are run. It is obviously possible to write this manually in the hook files, but I believe it would be great if this could be added directly into Mookme.

Concrete example I have a black formatter hook. When attempting to commit, black is run on the file, but the commit is not committed, as Mookme is set to exit on error. Now, parts of the file is staged, in an unacceptable state, and the parts that have been changed by black, are not staged. When attempting to commit anew, without first staging the new changes, the commit is accepted, as black is run against the working tree.

[LMaxence]

Hello !

You are pointing out a well known limitation on my side. Having an automatic stash in Mookme is something that I had wanted for a long time.

However there is one use case I couldn't crack so far, happening relatively often: If you stash something, make a change (for instance through a formatter), and unstash the changes, it becomes very likely to end up in merge conflicts.

As far as it looks like a "normal behaviour" to me, it will definitely look like a bug to many others.

If you have some ideas about the functional design of this feature, I would happily dig into it with you :)