Le mercredi 17 juin 2015, 15:24:56 Ken Thomases a écrit :
On Jun 17, 2015, at 2:59 PM, Andrew Eikum aeikum@codeweavers.com wrote:
I often try to write good commit messages, but it seems kind of pointless because I know they'll be removed.
They aren't universally removed. Alexandre sometimes edits them down, but most of the commits of my patches have a good chunk of what I wrote. And I'm often pretty verbose. (See, for example, http://source.winehq.org/git/wine.git/?a=commit;h=792b47ad3b89d51f17ef311c caa62fc2016af793.) It's always possible that I'm a special case because a) I mostly work on the Mac driver, about which Alexandre couldn't give two figs; and b) I've worn him down by writing too much too often. ;)
I try to help Alexandre by putting the core information that helps future developers (including future me) understand what the patch is for and what it does toward the top and then things like justification or help for reviewers lower down. That way, his edit just amounts to cutting the message off at a certain point.
It happens, that the first time I discovered that was with one of your changes :) I was trying to understand: https://bugs.winehq.org/show_bug.cgi?id=30984 which was impacting Gothic 2 for which I was making a new entry in the appdb.
The bug said it was a regression introduced by 43984f355a2905e16075a9df3d7fbe463761e853 is the first bad commit commit 43984f355a2905e16075a9df3d7fbe463761e853 Author: Ken Thomases ken@codeweavers.com Date: Thu May 31 15:33:23 2012 -0500
winex11: Be more conservative when matching keys from built-in layout tables.
The commit message is indeed terse :) After some time I found you're original mail which was much more explanatory. https://www.winehq.org/pipermail/wine-patches/2012-May/114666.html
I have no idea where Wine's policy of stripping commit messages comes from and I dislike it.
As I say, I don't believe there is such a policy. My impression is that you're not the only person who thinks there is, though. Between that and some developers' natural terseness (and, sometimes, assumption that their changes are always self-explanatory), few developers write much.
Another problem is that the email submission format doesn't offer a way for the submitter to separate text intended to be part of the permanent commit log from text intended to be merely commentary for consumption at review time. So, Alexandre has to exercise judgment about which parts are which kind and may mis-categorize.
-Ken
Isn't the text, that is after the --- and before the diff text, there for that purpose? I've tried to find the documentation, but I didn't succeed. And my git fu is not good enough..
Cheers, Marc