Enable cache on linting

[serl]

Problem description

We use both taplo-format and taplo-lint hooks in all our repositories. That's fantastic, but it feels a little slow.
Digging a little more, looks like taplo-lint fetches data from the internet on each run, and depending on the network, that adds a handful of seconds (or plain time-out error), which is bad for a pre-commit hook.

Proposed solution

taplo implements caching already!

After a mkdir .taplo-cache on my local repo, this works like a charm:

  - repo: https://github.com/ComPWA/taplo-pre-commit
    rev: v0.9.3
    hooks:
      - id: taplo-lint
        args: [--default-schema-catalogs, --cache-path=.taplo-cache]

While that works, there are some drawbacks:

• cache is not shared between projects (while I guess it's always the same)
• the .taplo-cache directory must exist already: taplo does not create by itself nor errors out when that's missing

Given that pre-commit creates a venv when running these hooks, it would be fantastic if the cache was over there.

What do you think? I'd like to create a PR if we agree on this issue. Also, if you have any suggestion don't hesitate.

What should the interface look like?

Additional context

No response

[redeboer]

After a mkdir .taplo-cache on my local repo, this works like a charm:

  - repo: https://github.com/ComPWA/taplo-pre-commit
    rev: v0.9.3
    hooks:
      - id: taplo-lint
        args: [--default-schema-catalogs, --cache-path=.taplo-cache]

Thanks a lot for sharing this trick!

Given that pre-commit creates a venv when running these hooks, it would be fantastic if the cache was over there.

What do you think? I'd like to create a PR if we agree on this issue. Also, if you have any suggestion don't hesitate.

That sounds like a great idea. I am not sure, however, whether hooks have access to the internals (caches) of pre-commit. At least, I can't find anything about this in the configuration of custom hooks.

Perhaps providing a 'patch' hook that creates that cache directory is an quick solution, but that still puts the caches locally in the repo.

[serl]

I was thinking that maybe we can wrap the lint call in a little python script, that should know where it lives. The rest should be straightforward. WDYT?

[serl]

I opened a PR with the idea: #29, seems to work locally. Let me know what you think.

I'd add a little more infrastructure (like pre-commit to run ruff&friends, including in CI - I mean it's a whole separate PR)

[redeboer]

Thanks a lot!
Yes there is currently no CI to keep this repo as small as possible for pre-commit. But we could add some CI if needed.