branch specific rules - the latest proposal

[Alexander Belchenko]

John Szakmeister пишет:
> On Fri, Nov 13, 2009 at 4:51 AM, Alexander Belchenko <bialix at ukr.net> wrote:
> [snip]
>> John, I want to say that I'm shocked and disappointed as well. But I knew
>> this fact (bzr will never support versioned properties) since April 2008
>> when Robert Collins rejected my naive attempt to introduce them in my
>> rejected naive patch for line-endings supports. I don't want to raise this
>> discussion again, sorry, it's still hurts for me. It's just good to know
>> that I'm not alone in my crazy wish for them.
> 
> You're definitely not alone.
> 
>> IIUC the "rules" introduced by Ian should be used instead. We just need to
>> change our minds and start using "rules". I think this is better than
>> nothing, so the faster the "rules" will materialize, the better for all of
>> us who need versioned properties.
> 
> Certainly.  I didn't mean to suggest otherwise.  And I, for one, am
> happy to see there will be a mechanism to propagate this data with the
> branch.  But if versioned properties are off the table, I'd at least
> like to find a way to interact with Subversion's properties so that we
> could have a better story on that front.  Jelmer is probably the best
> guy to speak about it, but it's one of the constant irritants that
> I've run into while trying to interact against Subversion
> repositories.


As I wrote there is problem with round-tripping. Definitely you can push 
properties from bzr to svn by converting .bzrmeta/rules on the fly, but 
the other way is problematic: you can't attach svn properties from past 
revisions to .bzrmeta/rules easily. Jelmer can clarify this point and 
correct me if I wrong.

Maybe (just maybe) it will be possible to use my idea with local 
settings, i.e. use .bzrmeta/rules.local to bring data from svn. But this 
will work only for the tip of the branch. But this is dirty ad-hoc.