Brian McGee

@michaelpj You could try the `--batch-size` arg by using this branch as a flake input. Reducing the batch size to `128` or `256` should further improve your use case. You...

@zimbatm if we deploy this though, the docs site will update with a feature that is not released... I'm currently looking into versioned docs with vitepress.

Paths to be formatted are batched according to a batch key unique to the sequence of formatters that matched against that path. Take the following config, for example: ```toml [formatter.foo]...

@michaelpj I did a quick spike [here](https://github.com/numtide/treefmt/pull/335/commits/f506357407434807d8993b1d3016effb90ab7515) so you can use this branch if you want to play around with configuring smaller batch sizes for `ormolu`. ```toml [formatter.ormolu] ... batch_size...

This is too stale, I'll re-implement on top of latest main

V2 has a new concept of walkers, with default performance greatly improved: ```console # nixpkgs with no cache traversed 42264 files emitted 42264 files for processing formatted 35462 files (623...

This would fall under an "ephemeral" system property and live behind the `--ephem` flag, meaning by default it wouldn't be captured.

@terlar trying upping the logging with `-v` or `-vv`, that might shed some light

Catching the failure and running `git diff` might shed some light too. If that shows nothing changed, then `treefmt` detects an incorrect change. It may be that a change is...

That's correct, v2 will not operate on untracked files when using the default `git` walker. If you use `--walk filesystem` it will just operate on whatever it finds.