[burntsushi]

I'm finding it difficult to follow the thread of discussion here, so I'll just blurt out some things:

* Most PRs that come in do indeed get squashed merged, but because of a few reasons. Sometimes it's because the PR is really just one logical change and really should just be one commit. Sometimes it's because the commit messages are not written in a way that I like, and so I do a squash-merge so I can either edit or re-write the commit messages provided by the author.

* Sometimes I do a "rebase and merge" if the commit breakdown is good and so are the messages. It's somewhat rare, but some contributors get it right. I'm fine with a single PR containing multiple commits, but the common case is indeed "one PR = one commit."

* I don't usually levy this criticism because I think it's low-brow, but the OP here---"git commit messages are useless"---is pretty egregious clickbait. Halfway into the article, the OP acknowledges that they aren't useless, but just useless in the context of a PR workflow. Which... I don't also 100% agree with, but is usually true for very small changes. What an annoying submission.

[thunky]

Thanks for weighing in.

I agree the title is clickbait but I didn't mind the article because it may challenge the reader to reconsider why they're putting so much effort into individual commits, even if they're not convinced to abandon (PR branch) commit messages entirely.