![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Commit attribution
Thinking about attribution. Right now, we have it so that the committer is logged as the patch author in mercurial, and the patch author is listed in the text of the commit message.
I wonder how much work it would be to have the author be in the mercurial log, and the committer noted in the text for accountability.
On the one hand, it would be easier to pull out a list of names for statistics! On the other hand, some commits need to be attributed to multiple people, and there's no way to log multiple authors for a patch other than just listing them all (which wouldn't help with statistics).
Eh, hm.
I wonder how much work it would be to have the author be in the mercurial log, and the committer noted in the text for accountability.
On the one hand, it would be easier to pull out a list of names for statistics! On the other hand, some commits need to be attributed to multiple people, and there's no way to log multiple authors for a patch other than just listing them all (which wouldn't help with statistics).
Eh, hm.
no subject
I could work on that within the confines of the current system, if you like.
no subject
This command counts the number of lines changed per user (I believe the random ones are my fault, from when I was doing hg qimport on a patch file that had author attribution on it, without doing a -u to override with my username). The churn-aliases file is just to map multiple emails/names to one in cases where there are slight differences but the person is the same:
$ hg churn --aliases .hg/churn-aliases
mark 921498 ***************************************************************
fu 184378 *************
kareila 29376 **
pauamma 2679
janinedog 1164
jproulx 58
jen@Nique 21
av8rmike 2
hg@creolibre 1
And it would be extra shiny to have the ability to look at http://hg.dwscoalition.org/dw-free/ and see who wrote the patch at a glance, not just who did the commit.
I think we can extract the person who did the change, just not as easily, but from there it should be easy to do everything else. Just (again) we wouldn't be able to take advantage of pre-coded stuff.
no subject
no subject
no subject
Once I finish parsing the backlog I'll put it up somewhere.
no subject
hg commit --user "Andrea Nall <[email protected]>" or whatever
I think the co-author thing could be solved by doing
Co-author: xxxxx
Co-author: xxxx
Committer: xxxxx
at the very top of the log message, and that way it'd be more machine parsable.
no subject
no subject