Whats the difference between check --fix and fmt

[jankatins]

The only description I found was Format Markdown files (always exits 0). (fmt) vs Lint and auto-fix issues (check --fix) (see https://rumdl.dev/usage/cli/#check-paths).

The pre-commit hook seems to use the latter, despite being called rumdl-fmt (https://github.com/rvben/rumdl-pre-commit/blob/32cfa95e49079b9925454b7ad805341d6bbc2839/.pre-commit-hooks.yaml#L9-L12).

Is there more of a difference between these two or is it really that both below are functional equivalent?

rumdl check --fix || true
rumdl fmt

[rvben]

The only real difference is exit codes. rumdl fmt always exits 0. rumdl check --fix exits 1 when something can't be auto-fixed. Same rules, same fixes, same file writes, just different behavior when violations are left over.

The rumdl-fmt pre-commit hook was quietly running rumdl check --fix under the hood so it would fail on unfixable issues. Made sense at the time, but the name didn't match what was actually running.

I just pushed a fix. The hooks now do what their names say: rumdl runs check --fix (fails if anything is left unfixable), rumdl-fmt runs rumdl fmt (pure formatter, always exits 0, fails only if files change). Same pattern as ruff's ruff + ruff-format split. Use both together, rumdl first.

If you were only using rumdl-fmt and want to keep catching unfixable violations, add the rumdl hook alongside it.

[jankatins]

Thanks!

I now use fmt (outside of precommit) before check (in precommit, which also runs in the CI Pipeline).

I added #553 to make the (non-)difference clearer.