Re: [PATCH 1/2] t3200-branch: test setting branch as own upstream

From: Jeff King <peff@peff.net>
Date: Feb 28, 2014, 13:03 UTC
Message-ID: <20140228130340.GB23448@sigill.intra.peff.net>
In-Reply-To: <CAN7MxmVxYhF1VXWZNAXyfF8iBNQaqqwoyjmDAP_9_bc5NUUU6A@mail.gmail.com>

On Fri, Feb 28, 2014 at 08:16:13PM +0900, Brian Gesiak wrote:

Show 7 quoted lines

> > I just don't want to regress somebody else's workflow due
> > to my lack of imagination.
> 
> This makes a lot of sense to me, although as-is the function emits a
> warning and returns immediately (although with a successful status
> code), so I'm also stumped as to what kind of workflow this would be
> included in.

I'm cc-ing Matthieu, who wrote 85e2233, which introduces the check. Its commit message says:

  branch: warn and refuse to set a branch as a tracking branch of
  itself.

  Previous patch allows commands like "git branch --set-upstream foo
  foo", which doesn't make much sense. Warn the user and don't change
  the configuration in this case. Don't die to let the caller finish its
  job in such case.

For those just joining us, we are focused on the final statement, and deciding whether we should die() in this case. But I am not clear what job it would want to be finishing (creating the branch, I guess, but you cannot be doing anything useful, since by definition the branch already exists and you are not changing its tip). There wasn't any relevant discussion on the list[1]. Matthieu, can you remember anything else that led to that decision?

> In any case, if the jury's out on this one, I suppose the two patches
> I submitted are good to merge? Part of me thinks the bump from warning
> to error belongs in its own patch anyway.

Yeah, it should definitely be a separate patch on top.

-Peff

[1] Threads are:

      http://thread.gmane.org/gmane.comp.version-control.git/137297/focus=137299

and

      http://thread.gmane.org/gmane.comp.version-control.git/137401/focus=137403

    but the discussion focused on the first part of the series.