histedit + rollback destroys working copy

[Colin Caughie]

> -----Original Message-----
> From: mercurial-devel-bounces at selenic.com [mailto:mercurial-devel-
> bounces at selenic.com] On Behalf Of Augie Fackler
> Sent: 02 March 2011 15:22
> To: Dominik Psenner
> Cc: Mercurial Developers
> Subject: Re: histedit + rollback destroys working copy
>
> On Wed, Mar 2, 2011 at 2:57 AM, Dominik Psenner <dpsenner at gmail.com>
> wrote:
> > Hi,
> >
> > I most probably was able to produce a workflow that nukes a
> repository
> > because hg keeps complaining about "unknown parent revision"
> without letting
> > me do anything to change that. Whatever I want to do on the
> working copy
> > (up, shelve, ..), hg keeps complaining:
> >
> > $ hg up null
> > Abbruch: Arbeitsverzeichnis hat unbekannte Vaterversion
> '7153d5bf1703'!
> >
> > So, what I did was:
> >
> > * at "tip-2" I committed a partial fix
> > * at "tip-1" I committed the second part of the fix
> > * then I told histedit to fold tip and "tip-1" into one commit
> > * now the previous "tip-2" became a new changeset too
> >        => this could have been my fault, I am unsure if I marked
> "tip-1" or
> > "tip" to fold
> > * at this point I started a rollback which removed the changesets
> tip,
> > "tip-1" and the new "tip-2" leaving my local changes intact
> > * and here I am with a working copy that has a unknown parent
> >
> > Is someone able to reproduce this?
>
> I'll take a look.

One of my colleagues did something very similar the other day, but using the collapse extension rather than histedit, and got exactly the same result. So I suspect it's a general problem with using rollback after a command that modifies history.

If you figure out a way to get histedit to avoid this issue, please let me know and I'll fix up collapse as well.

Regards,
Colin


Latest News at: http://www.indigovision.com/news2011.php