[PATCH 5 of 9 phases] phase: add mechanism to store phase data into dirstate

[Pierre-Yves David]

On Thu, Jan 05, 2012 at 10:09:28PM -0600, Matt Mackall wrote:
> On Thu, 2012-01-05 at 02:25 +0100, Pierre-Yves David wrote:
> > # HG changeset patch
> > # User Pierre-Yves David <pierre-yves.david at ens-lyon.org>
> > # Date 1325725382 -3600
> > # Node ID 9bedf170a2c4753b01296db5f8dc8141cd3a64ed
> > # Parent  6b460b7a2e2a2788031c4c5e8dd37160ddbd7ad8
> > phase: add mechanism to store phase data into dirstate
> 
> (store IN)
> 
> Ok, what happens when I'm working on my secret branch and then switch to
> the development tip and commit?

The phase data register through the phase command is lost on update

> > +        When force is True, the next commit is required to be exactly in the
> > +        specified phase.
> 
> ???
> 
> I really don't like the idea of stashing away a force switch for an
> indefinite period of time.

I was looking for a way to say "I want the next commit to be un phase X
whatever parents phase were".

Do we need another way to implement this usecase or shall it be ignored ?

> > +        """
> > +        try:
> > +            content = self._opener.read("phase").strip()
> > +        except IOError:
> 
> We allow all IOErrors?

Same code than branch one. I'll send a patch for it too.

-- 
Pierre-Yves