DragonFly commits List (threaded) for 2008-12
Re: commit mail subject format
Peter Avalos schrieb:
On Sun, Dec 07, 2008 at 03:15:41PM +0100, Sascha Wildner wrote:
After replacing the paths with summaries, how would someone search for
commits to certain files using only his mail/news client?
Ugh. Why would they do that? I don't think this is something we should
try to support when git does a MUCH better job.
Full text search? Hmm, not possible using NNTP unless you download all
articles. Use gitweb or search the repo? Well not all of us read the
mailing lists on the same box (or platform even) where they keep the
repo. And having to keep an additional window open just to see what was
changed doesn't seem like an improvement for me, personally.
But your argument was for people to search for commits to certain files
using a mail client. This seems really silly to me.
*sigh* That seems to be your (narrow, perhaps) view on different
workflows or other people's demands. I'm fine with agreeing to disagree.
I was just pointing out what works now and won't work with summaries as
Summaries are sub-optimal for the Subject since they are human generated
and humans make spelling errors, might forget the
commit-structuring-rules and tend to formulate things differently from
what the guy who is searching might expect. I can understand that it
sounds like a good thing, but I fear that daily practice will not live
up to it.
This is easy though. Hell, it's already mentioned in committer(7).
I know that it's mentioned there.
But mentioning something in committer(7) doesn't prevent people from
doing spelling errors or forgetting the first-sentence-rule anyways or
writing stuff that doesn't give you a clue etc. etc.
And it just seems silly to me to rely on all committers' abilities to
always think up a meaningful one-sentence summary instead of something
which is auto-generated and (while not optimal either) is at least reliable.
Go and take a look at past commits which had one sentence as message.
You'll find that the average committer is used to assuming that people
know the location of the change:
swildner: Improve markup.
sephe: Deprecate MALLOC/FREE macros
sephe: Correct the reversed comparison logic
hasso: Make CIDR in "-net net gw mask" form work again.
In my opinion, the sentences are meaningless (even misleading) if taken
alone. And if I have to open the mail anyways to get the picture, then
what use is the Subject at all?
My criterion for deciding which commit mails interest me immediately is
mainly the location of the change. So naturally, I find the path in the
subject much more helpful.
Just my 2c,