mq and named queues - and qseries --edit. Was: Re: mercurial at 11238: 10 new changesets (3 stable)

[Martin Geisler]

Gilles Moris <gilles.moris at free.fr> writes:

> On Sunday 06 June 2010 04:51:26 pm Martin Geisler wrote:
>> Kevin Bullock <kbullock+mercurial at ringworld.org> writes:
>> > On 4 Jun 2010, at 1:58 PM, Martin Geisler wrote:
>> >> The idea is that instead of editing the series file manually, I would do
>> >> something like
>> >>
>> >>  hg qpush --grab foo
>> >>
>> >> to apply patch 'foo' next, thereby editing the series file on the fly.
>> >> This would basically turn the unapplied patches into a *set* instead of
>> >> a *list*, allowing you to grab any unapplied patch and apply it.
>> >>
>> >> I would even consider making --grab the default when there is a named
>> >> patch on the command line. That way
>> >>
>> >>  hg qpush     <-- apply next patch in series
>> >>  hg qpush foo <-- make 'foo' next and apply it
>> >>
>> >> You would then use
>> >>
>> >>  hg qgoto foo
>> >>
>> >> to get the old behavior of
>> >>
>> >>  hg qpush foo
>> >
>> > FWIW I agree with Mads that this would be dangerous.
>>
>> Yes, reordering patches are of course dangerous.
>>
>> > I like the idea of having a --grab option, but not making that
>> > behavior the default.
>>
>> No, that would probably be too much, not to mention totally backwards
>> incompatible.
>>
>> > Here's why I like the option: the series file, although it's in
>> > plaintext, is essentially serialized state of the patch queue. Letting
>> > the software manipulate it that knows about its format seems
>> > preferable in the normal case, even if manual hacking ability is nice
>> > to have.
>>
>> Right.
>
> Isn't qpush --move already that ?

Ehm, yes it is... how strange that I hadn't seen that option sneak in!

-- 
Martin Geisler

aragost Trifork
Professional Mercurial support
http://aragost.com/mercurial/