Re: [PATCH] Try harder to find a remote when on a detached HEAD or non-tracking branch.

From: Heiko Voigt <hvoigt@hvoigt.net>
Date: Jun 20, 2012, 17:42 UTC
Message-ID: <20120620174213.GA2500@book.hvoigt.net>
In-Reply-To: <4FE0F262.9050404@xiplink.com>

Hi,

On Tue, Jun 19, 2012 at 05:42:58PM -0400, Marc Branchaud wrote:

Show 7 quoted lines

> On 12-06-19 03:31 PM, Heiko Voigt wrote:
> > Using the first configured remote as a fallback also sounds quite
> > arbitrary to me. Your current HEAD might have nothing to do with that
> > remote.
> 
> Sure, but if it doesn't specify a remote then why not?  (Instead of "first
> configured remote" think "remote used in the initial clone command".)

Returning the "remote used in the initial clone" sounds ok to me as well but that does not have to be the same as "first remote git config --list returns".

Show 10 quoted lines

> > Before falling back to origin how about guessing the remote from the
> > first branch thats contained in HEAD?
> >
> > To me that sounds quite natural. The first branch could be ambiguous so
> > we would have to come up with some ordering. Maybe a depth search and
> > first parents first? Or a breadth first search with younger generations
> > first and then first parents first?
> 
> This sounds much harder to explain to a user than "the remote you used when
> you cloned the repo".

Well I think "it uses the remote of the branch that you based your work on" is not hard to explain. What's harder is the implementation.

> > Would that work for your use case Marc?
> 
> Maybe, but it seems much more complicated than necessary.

Git is a tool used in very different workflows so a change needs to be quite generic. Make it as simple as possible but *no simpler*. I think your patch is currently not respecting the "no simpler".

Cheers Heiko