Request for rebaseif extension to be provided by default with rebase

[Sébastien Deleuze]

Thanks for your feedback Matt.

About fetch I agree with you, it should not be used.

The idea behind my request is more to allow users to "secure" existing
"hg rebase" and "hg pull --rebase" command, not to create a new
shortcut for people coming from CVS/SVN world.

Instead of creating a new command, we perhaps could just add a
preference in .hgrc file in order to control "hg rebase" and "hg pull
--rebase" behaviour in order to let people decide what should occur
when there is conflict during a rebase.

[rebase]
onconflict=interrupt (default), abort or abortAndMerge

- interrupt : interrupt and let the repository in a REBASING state
(current behaviour)
- abort : abort (behaviour I prefer). I case of hg pull --rebase, it
means that in case of rebase conflict, it will be equivalent to a
simple hg pull
- abortAndMerge : abort and run a merge (current rebaseif behaviour)

And maybe some command ligne option like hg rebase --abortonconflict

Hope it will make sense for some expert users, and/or that
implementing it as configuration capabilities instead of a new command
will convince you ;-)

Regards,
Sébastien

2011/5/24 Matt Mackall <mpm at selenic.com>:
> On Tue, 2011-05-24 at 17:51 +0200, Sébastien Deleuze wrote:
>> Could we have a feedback from Mercurial core developers (since rebase
>> is a core extension) in order to know if a patch that would add
>> rebaseif functionnality to rebase extension, perhaps with minor
>> improvements like with internal:fail behaviour, would be reviewed if
>> sent on mercurial-devel at selenic.com ?
>
> I'm not terribly keen on the idea. It's a bit too reminiscent of fetch,
> which seems to mostly exist so that CVS refugees can put off actually
> understanding Mercurial long enough to make really big mistakes rather
> than small ones.
>
> One measure of this is that no expert users care about fetch in the
> slightest. It's just not part of their workflows or the workflows they
> recommend to their users. As such, the code has been neglected for
> years.
>
> I fear we'd have a similar situation with rebaseif: lots of new users
> using it, coming to the experts for help, and the experts saying "you
> shouldn't be using that".
>
> But if some expert users want to weigh in in favor of rebaseif, I'm
> listening.
>
> --
> Mathematics is the supreme nostalgia of our time.
>
>
>