Joshua D. Drake wrote:
> On 10/13/2015 08:15 AM, Tom Lane wrote:
> >Andres Freund <andres@anarazel.de> writes:
> >>On 2015-10-13 16:21:54 +0200, �lvaro Hern�ndez Tortosa wrote:
> >>>(50 chars for the commit summary, 72 chars line wrapping)
> >
> >>-1 - imo 50 chars too often makes the commit summary too unspecific,
> >>requiring to read much more.
> >
> >I agree --- I have a hard enough time writing a good summary in 75
> >characters. 50 would be awful.
>
> The idea of writing a commit message that is useful in a number of
> characters that is less than half a tweet sounds unbearable. The idea of
> trying to discern what the hell a commit actually is in a number of
> characters that is less than half a tweet sounds completely ridiculous.
There are many commits that can be summarized in small lines; new
features are often like that. Bug fix summaries are much harder to
write and most of the time they require longer lines. When you can
achieve 50-char lines it looks better in tooling (gitweb or git log
--oneline), but if you have to make it 75 it's not the end of the world.
http://git.postgresql.org/gitweb/?p=postgresql.git;a=shortlog
I don't care one bit if one or two lines that are part of the _body_ of
the commit messages are longer than 80 chars, particularly if they
provide useful links such as message-ids. The message-ids or
message-id-based URLs are too handy to ignore.
A month ago I asked sysadmins@pg.org about using
http://postgr.es/m/<message-id> as a way to create shorter URLs to use
in commit messages, but the idea wasn't too hotly received so I let it
go. I'm glad it popped up again.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
