[PATCH 3 of 3] histedit: test that an aborted histedit can be rerun (with obsolete)

[Augie Fackler]

On Mon, Mar 27, 2017 at 02:04:44PM +0200, Pierre-Yves David wrote:
>
>
> On 03/27/2017 12:31 PM, Jun Wu wrote:
> > Excerpts from Pierre-Yves David's message of 2017-03-27 09:19:52 +0200:
> > >
> > > On 03/26/2017 08:42 PM, Jun Wu wrote:
> > > > Excerpts from Jun Wu's message of 2017-03-26 09:59:43 -0700:
> > > > > I'm -1 on this series.
> > > > >
> > > > > If the goal is to workaround obsolete cycles, the code could do additional
> > > > > checks "if the destination is obsoleted" and do "hg touch"-like thing
> > > > > automatically.
> > > >
> > > > FYI I have sent the series doing the above.
> > >
> > > I've commented on that series too, your new behavior still introduce
> > > issue in distributed case.
> > >
> > > I think we should backout for now (since this is a regression, that
> > > exist on default right now).
> > > In the mean time we take a step back and think about the situation cold
> > > headed.
> > >
> > > See comment on the other series for details,
> > >
> > > Cheers,
> > >
> >
> > I have resolved the issue you commented at [1].
>
> Actually no there are still issues. Let us channel the discussion here,
> spreading the discussion across multiple threads is getting confusing.
> Especially when series rely on other unlanded series.
>
> > I'm still not convinced that we have to go back to use strip.
>
> So far, we have a regression at hand and no known path forward fully solving
> the issue. I would like you to slow down and discuss possible alternative to
> backout in this thread. Please to not rush to writing code and do not send a
> new series until we reached a consensus here on the problem space and a
> viable solution.

Pierre-Yves is right: this is a regression, and needs to be fixed
before 4.2 is final. Given that the freeze is in three weeks, I'm
going to take this series and if we can work out a more elegant fix in
the next couple of weeks I'll be very happy.

In general, regressions are The Worst. We've got a few open regression
bugs in the BTS that have been open unacceptably long, and I'm
inclined to be aggressive about avoiding any more cases like that.

(Jun: don't take my acceptance of these changes as a rejection of your
ideas! It's just temporary damage control until we get a better
solution in place.)

>
> Cheers,
>
> --
> Pierre-Yves David
> _______________________________________________
> Mercurial-devel mailing list
> Mercurial-devel at mercurial-scm.org
> https://www.mercurial-scm.org/mailman/listinfo/mercurial-devel