Re: Avoiding 'master' nomenclature

From: Michal Suchánek <msuchanek@suse.de>
Date: Jul 30, 2020, 10:11 UTC
Message-ID: <20200730101139.GB32107@kitsune.suse.cz>
In-Reply-To: <CAHk-=wjN0o-+OAn5wXMcshLGQ4H+CLJwq8+wRnaqzKMuhbmZqw@mail.gmail.com>

On Wed, Jul 29, 2020 at 05:23:59PM -0700, Linus Torvalds wrote:

Show 20 quoted lines

> On Wed, Jul 29, 2020 at 5:14 PM Jeff King <peff@peff.net> wrote:
> >
> > I'm on the fence on how magical to make the default. Having "master"
> > there gets Linus's case back where he wanted without having to configure
> > anything, which is probably reasonable. I'm not sure if people would
> > want their init.defaultBranch in addition / instead.
> 
> Junio seemed to go for "instead", but I think it might be more natural
> to just have "master" as the initial entry, and anybody adding entries
> will add it to the list.
> 
> I do think it might be a good idea to make "git init" just add the
> entry from whatever the default initial branch is.
> 
> And then the "empty entry to clear" can be used to _force_ a clean
> slate, although I don't see why anybody would ever really want that.
> If you make your default branch name be "develop", and you really want
> to see the "into develop", you'd remove the entry that "git init"
> would hypothetically add, you wouldn't necessarily want to do a "clear
> list".

Please don't. I have seen this in systemd services and it is a mess to deal with from user side. I suppose it is a mess on the implementation side as well. This mystery variable content is just pain and the complexity of the problem at hand does not really call for such mostrosity.

FTR I don't really care about the omission of target branch in merge messages. I typically work with repositories that are about three levels of indirection away from the 'master' branch that would get omitted by deafult. But in the systemd case you do need to care to expunge the previous broken values and it is a real pain.

Thanks

Michal