On Thu, Apr 01, 2021 at 10:25:50PM +0200, Stefan Sperling wrote:
> I don't see any errors in your patch. However, for the main branch,
> I would rather have a diff (or a series of diffs) which supports two paths
> as well as two object arguments, and supports an option similar to
> 'got cat -P' to handle ambiguous cases, and adds regression tests or
> tweaks existing tests to cover new or changed behaviour.
Sure, count this as the first diff amongst more.

> The reason is simply that I wouldn't consider this intermediate state
> ready to be released. If something else important comes up out of the
> blue I'd rather not be blocked from releasing a new version immediately
> because an incomplete feature implementation sits on the main branch.
I agree that this half-arsed state shouldn't land in release.

> That said, there's no reason to delete your diff. It is a good start and
> I'd be happy to keep reviewing additional diffs that apply on top of it,
> even in case you yourself feel that it's not quite ready and would like
> feedback to help with making progress.
Cool, I'll just post follow-up diffs to this thread.

> I've certainly been guilty of committing work-in-progress stuff in the past.
> These days I try to keep my development state on local branches and commit
> to the main repo only if I am confident that the current state could be
> shipped in a release (my recent deltify commits for example were originally
> a much longer local series of patches).
I'm working in a feature branch as well, so I'll keep doing that with
periodic rebases onto main.

Thanks!