Carlos R. Mafra wrote:
> On Sun 9.Nov'08 at 23:43:48 +0100, Carlos R. Mafra wrote:
> > On Sun 9.Nov'08 at 14:00:42 -0800, John H. Robinson, IV wrote:
> > > Carlos R. Mafra wrote:
> > > > On Sun 9.Nov'08 at 12:28:53 -0800, John H. Robinson, IV wrote:
> > > > > Carlos R. Mafra wrote:
> [....]
>
> > If the end result is to make more sense to everybody, "user" should be
> > used as the author.
>
> It is not just me who thinks like this, btw.
>
> Today there was an email in the git list from someone trying
> to import patches from Mercurial into Git, and one of the
> lines from his code is
Git has separated the idea of ``Author'' and ``Committer''. CVS, RCS,
Subversion and Mercurial don't. I am certain there are other systems
that also don't separate the concept. Git does separate it. I am not
surprised that those that are used to Git's separation become confused
when the roles are not separated.
In the common case, author and committer are the same. In some other
cases, they are not. The question is what to do when committer !=
author.
In my opinion, in CVS's opinion, and Mercurial's opinion that Mercurial's
label of ``user'' is that of committer.
% hg help commit | grep user
-u --user record user as committer
Accordingly, I contend that most mercurial users think of user as
committer, and if we try to go against that we would end up fostering
far more confusion than by keeping user as committer.
> author = re.search("# User (.+)", p).groups()[0]
>
> That is what makes sense to me.
Coming from Git, this does not surprise me. Coming from CVS, SVN, RCS,
or a plethora of other version control systems, user is committer.
> The complete email is here:
>
> http://article.gmane.org/gmane.comp.version-control.git/100636
I don't read python too well. It looks like it takes Mercurial's
username field and applies it to author, then uses Git's default idea
of committer as committer.
It sounds like a reasonable solution. As long as the mercurial log
indicates the correct information - who authored what parts - everything
should be fine.
When taking from a non-git source, Git has to add information. When
going from Git to non-git, then information has to be lost - or moved
into the free-form commit log.
I propose that we agree that Git and Mercurial are not the same, that we
agree that Git's idea of committer and Mercurial's idea of username are
analogous, and that Git's author header be incorporated in the free-form
commit log.
This will give us the benefit of being able to attach the correct author
to the correct verbiage in the changelog in the case that a particular
commit has parts from more than one author.
--
John H. Robinson, IV [EMAIL PROTECTED]
http ((((
WARNING: I cannot be held responsible for the above, sbih.org ( )(:[
as apparently my cats have learned how to type. spiders.html ((((
--
To unsubscribe, send mail to [EMAIL PROTECTED]
