On Tue, Dec 14, 2021 at 12:47 PM Mario Carneiro <[email protected]> wrote:

> One thing that I think we should stick to is one commit per PR. We had a
> similar discussion for metamath-knife but for set.mm it's important to
> keep PR sizes down so that people can reasonably review them, and squash
> merges keep things organized. This is the mechanism we've been using in
> mathlib and it works well to avoid merge conflicts, especially when things
> scale to many concurrent PRs.
>

For big PRs that benefit from multiple commits, I think the commits should
be consciously organized, squashing away bugfix commits and rebasing (not
merging) on master. Such PRs are merged using the "rebase" option instead
of "squash", but "squash" should be the default (and "merge" should not be
used).

-- 
You received this message because you are subscribed to the Google Groups 
"Metamath" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/metamath/CAFXXJSsDeUTNiUSskCMh6Gzuk5ZvqG29LkqPQV-RiS9nMQ04RQ%40mail.gmail.com.

Reply via email to