From: Omar Polo <op@omarpolo.com>
Subject: Re: rebase suffers '?' conflicts
To: Stefan Sperling <stsp@stsp.name>
Cc: Mark Jamsek <mark@jamsek.com>, Christian Weisgerber <naddy@mips.inka.de>, gameoftrees@openbsd.org
Date: Wed, 12 Apr 2023 20:31:16 +0200

On 2023/04/12 17:44:28 +0200, Stefan Sperling <stsp@stsp.name> wrote:
> On Wed, Apr 12, 2023 at 05:09:07PM +0200, Stefan Sperling wrote:
> > On Thu, Apr 13, 2023 at 12:21:29AM +1000, Mark Jamsek wrote:
> > > Nice! Thanks for the explanation, it all makes sense and is good to
> > > know. I thought this was going to be a lot more work to fix.
> > 
> > My fix is not quite correct. The contents of the file 'beta' are
> > still from newbranch at the end of the test.
> > I need to take another look at this.
> 
> Turns out there is another fairly trivial bug:
> 
> rebase -a is opening the wrong commit to diff against for checkout.
> Files should be checked out from the commit we are moving the work
> tree towards, rather than from the tip commit of the temporary branch
> we are moving away from.
> To fix the issue we can open the commit once the base ID has been
> changed, instead of before. Now we see beta getting updated as part
> of 'got rebase -a', as it should be.
> 
> The histedit abort command has the same bug.
> 
> The previously diagnosed issue is real regardless and affects a regression
> test in update.sh. I have tweaked this fix slightly to only set the new
> base commit ID if one is expected to be set on this file index entry.
> 
> Two diffs below:

ok op@