On Wed, Oct 12, 2022 at 10:21:39AM -0400, Theodore Ts'o wrote:
On Wed, Oct 12, 2022 at 02:16:42PM +0100, Luís Henriques wrote:
Grr, looks like I accidentally reused a 'git send-email' from shell history which had a '--in-reply-to' in it. Please ignore and sorry about that. I've just resent a new email.
No worries! The --in-reply-to wasn't actually a problem, since b4 generally will do the right thing (and sometimes humans prefer the in-reply-to since they can more easily see the patch that it is replacing/obsoleting).
b4 can sometimes get confused when a patch series gets split, and both parts of the patch series are in a reply-to mail thread to the original patch series, since if it can't use the -vn+1 hueristic or the "subject line has stayed the same but has a newer date" hueristic, it falls back to "latest patch in the mail thread". So if there are two "valid" patches or patch series in an e-mail thread, b4 -c (--check-newer-revisions) can get confused. But even in that case, that it's more a minor annoyance than anything else.
So in the future, don't feel that you need to resend a patch if there's an incorrect/older --in-reply-to; it's not a big deal.
Great, I haven't yet included b4 in my workflow so, to be honest, I didn't really thought about that tool being confused. What really made me resend the patch was that I used the *wrong message-ID in the "--in-reply-to"! And that thread already had a v2 patch, which would could easily confuse humans. Hopefully, b4 won't be confused by that either.
Cheers, -- Luís