Re: [StGit PATCH 1/2] Convert "stg refresh" to the new infrastructure

From: Karl Hasselström <kha@treskal.com>
Date: Jun 29, 2008, 13:46 UTC
Message-ID: <20080629134654.GB5098@diana.vm.bytemark.co.uk>
In-Reply-To: <b0943d9e0806290407h5eecd27bg510dd09e0188abca@mail.gmail.com>

On 2008-06-29 12:07:42 +0100, Catalin Marinas wrote:

Show 13 quoted lines

> 2008/6/29 Karl Hasselström <kha@treskal.com>:
>
> > On 2008-06-29 10:42:32 +0100, Catalin Marinas wrote:
> >
> > > Does it make sense to refresh an unapplied patch? Maybe adding a
> > > new file to the patch but I don't really see a need for this.
> >
> > But no, I don't have a strong sense that this is super useful. It
> > was just easy to allow, so I allowed it.
>
> It seems harmless, unless someone finds some unusual behaviour. What
> is the conflict behaviour? Is the refresh aborted? For unapplied
> patches, it is more complicated to let the user solve the conflict.

First, the refresh is stored in a temp patch. Then, that patch is popped. Then we try to coalesce the temp patch and the target patch; if this fails due to conflicts that we can't represent because we have no index and worktree to work with for unapplied patches, we just leave the temp patch in the stack for the user to deal with. (E.g. by trying to apply the two patches and sort out the conflicts, by keeping the temp patch separate, or by coalescing it with a different patch.)

-- 
Karl Hasselström, kha@treskal.com
      www.treskal.com/kalle