Peter Eisentraut <pete...@gmx.net> writes: > On 04/27/2016 11:04 PM, Tom Lane wrote: >> Actually, git_changelog can merge identically-messaged commits despite >> intervening commits. It's set up to not merge commits more than 24 hours >> apart, though. We could loosen that requirement but I'm afraid it would >> make things confusing to merge far-apart commits.
> ISTM that git_changelog should be looking at the AuthorDate instead of > the CommitDate. Then it would work correctly for backpatches done using > cherry-pick. Meh. That would also make it noticeably *less* accurate for some other scenarios. It actually did look at AuthorDate to start with, and we changed it because we didn't like the results; cf 7299778a9. Also, IME, backpatching with cherry-pick fails often enough that designing one's process around it is just asking for pain. Certainly, many fewer than half of my own back-patches manage to use cherry-pick. regards, tom lane -- Sent via pgsql-hackers mailing list (email@example.com) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers