Thoughts on git commit mail
Ulrich Spörlein
uqs at freebsd.org
Fri Jan 1 11:58:56 UTC 2021
On Thu, 2020-12-31 at 19:03:37 -0500, Ed Maste wrote:
>On Thu, 31 Dec 2020 at 17:14, John Baldwin <jhb at freebsd.org> wrote:
>>
>> The mix of fields for name and date is a bit odd though. Perhaps another
>> alternative might be to elide the Author fields if the Author == Committer?
>> That would keep a lean format when when the individuals are the same.
>> For that suggestion, I think you would always show the two Commit fields
>> ("Committer" and "CommitDate") and only show "Author" and "AuthorDate"
>> if Author != Committer?
>
>I discussed this a bit with John on IRC, but for the benefit of the list:
>
>We expect Author == Committer will be common, and the commit date is
>probably the more interesting date. In addition, committer will always
>be someone at FreeBSD.org, so what about if we take John's suggestion for
>Date, and use the author name, appending (via /committer/) if it's not
>the author?
>
>Thus, for the example above:
>
>Author: John Baldwin <jhb at FreeBSD.org> (via rlibby)
>Date: 2020-12-31 19:56:33 +0000
>
>This is nice and concise while conveying all useful information, IMO.
>This would work identically for MFCs of commits made by other
>committers, as well as committing changes submitted by 3rd parties.
I have no strong preference either way (because I use git log if I want
to find out about things, something that now everyone has downloaded and
is fast, as compared to SVN, so maybe behaviors need to change a
little.)
Should we drop Committer and CommitDate from the logs entirely, because
99% of the time it's just what is in the commit email From: and Date:
headers anyway?
The thing is, folks that take their information from commit emails see
the Date of the push and usually who pushed it (Committer) in their
email client. The really nice information on these MFCs is, how old
their actual author date is, as that shows you how laggy we are with
MFC'ing.
And for commits to head, it's super interesting to see the potential 3rd
party Authors we'll see in the future, and, again, the committer and the
push date is shown in the email headers, so it's kinda redundant.
I would be biased against a more dynamic format that only sometimes
shows information and sometimes doesn't, as that can be confusing.
Cheers
Uli
More information about the freebsd-git
mailing list