[darcs-users] Subtitution of dependencies (was: darcs conflicts/dependencies )

[Eric Kow]

Thanks,

On 13 September 2012 15:30, Gracjan Polak <gracjanpolak at gmail.com> wrote:
>> If so, that is not normal.  The behaviour I would expect is
>>
>> - it offers B first
>> - if you reject B, it does not offer S
>
>
> Hmm... in darcs 2.4 if I remember correctly darcs autoincluded patches that
> were dependencies. So it asked about patch [1/2], then knew that [2/2] has
> to be in because of dependencies and darcs did not ask about it.

No, darcs interactive mode is conservative/passive in its approach, the idea
being that it only skips things that you don't give it the
dependencies for.  But unless
I'm missing something, you're using a flag that overrides this passivity
(which you weren't), it won't go out of its way to fulfil dependencies for you.

If you were using an operation that works from reverse, like darcs obliterate,
then it would ask you if you wanted to obliterate S.  If you said no (you
wanted to keep it), it would not bother to offer you B because it can't do it
without violating the dependency.

It's also possible that you might be remembering these unexpected dependencies
from a different part of the user interface, particularly the --match
or --patch options.
There, it can arise that you ask darcs something like

   darcs pull --match 'author Eric'

and darcs offers me patches from Ganesh.  See
http://darcs.net/FAQ#darcs-offers-me-unexpected-match-results

Thanks for the follow-up!


-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9