[PATCH 3 of 5] subrepo: fix hgrc paths section during subrepo pulling

[Edouard Gomez]

On Mon, 15 Mar 2010 11:46:47 +0100, Saint Germain wrote:

> Hello,
> 
> 2010/3/15 Edouard Gomez <ed.gomez at free.fr>:
> 
>>> For my part I would very much like any input/advice about this patch:
>>> http://www.selenic.com/pipermail/mercurial-devel/2010-
March/019313.html
>>> As I personnaly think that it is essential to subrepo management (to
>>> guarantee that we are in a particular state with no 'dirty' subrepo).
>>
>> Should be nice to force a clean state on subrepos too when main is "up
>> -C"
>>
>> However I would prefer to have default behavior to recurse and a "-N"
>> option for a not recursive behavior.
>>
>>
> Steve Losh (sjl) was (correctly I think) afraid that a "up -C" would
> carelessly wipe out any changes not commited on the subrepo. So we
> decided on an additional flag to specify that we _really_ want this
> behavior and an error message if we just try to use "up -C". However 
> this might be a good idea to indroduce a --subrepo flag which could be
> also used for any operations that might also operate on subrepo (like
> "status" for instance, or even "commit").
> 
> What do you think about it ?

Basically i'd say that subrepositories should not break the
"atomic" nature of a mercurial repo. I expect a repo, to always
behave as a whole.

So that would mean for me:
 - diff, and stat should recurse in subrepos, because the commit operation
   will validate the whole set of repos as a whole, so reviewing all
   the contents is normal and expected.
 - push/pull must recurse into subrepos. Mandatory because if that's not
   the case, it'd be easy to end up with main repos lacking their
   dependencies.
 - "up" should not recurse because i think that's a developper choice
   to use a specific version of a sub project (implies merge, testing...)
 - "up -C" should init all subrepos back to their .hgsubstate state.
   (your patch)
 - commit should recurse into subrepos whatever their deepness (my patch)

Now it's true that having "up -C" behaving differently from 'up' is hard
to defend. It just seems more natural to me.

But I think that we will eventually have to add a global option so that
recursing is on or off, and can be setup by the user using the [defaults]
hgrc mechanism if the default behavior doesn't match their expectations.

Shaping default behavior being 2 working with subrepos is still better
than being alone :-) If more users come up, we could start a poll.

-- 
Edouard Gomez