[RFC] branch --bind

[Scott Aubrey]

On 8 Jan 2010, at 10:38, Vincent Ladeuil wrote:

>>>>>> "robert" == Robert Collins <robertc at robertcollins.net> writes:
> 
>    robert> On Fri, 2010-01-08 at 09:55 +0000, Scott Aubrey wrote:
>>> Also, forgot to add to the short term list that I'm in favour of adding --bind to branch too, for the same reason as setting bzr bind to use parent if no previous bind location is specified.
> 
>    robert> I recall Aaron having concerns about 'bzr bind' using the parent
>    robert> location, and that itself also lends concern to 'bzr branch --bind'
>    robert> binding to the parent.
> 
> *I* would very rarely bind to the *parent* because I'd like to
> have the proposed changes reviewed before being merged to the
> said parent.
> 
> In other words: parent is where I read from, bind target is where
> I publish my changes. For some workflows they are the same.
> 
> And that is my main concern with --bind: it binds to the parent
> (without a way to change that decision) and that's not my main
> use case.
> 
> So I'm far more likely to use 'bzr bind :push'.
> 
> There are cases where I have parent_location == push_location,
> but most of the time I bind after pushing so ':push' is *always*
> right.
> 
> But since I obviously 'bzr push' long after 'bzr branch' and that
> in most of the cases I don't push to the parent, I doubt I will
> ever use 'bzr branch --bind'.
> 
>  Vincent
> 
That's fine, and I agree completely... but what if your the one doing the landing after reviewing? Can you not see a case for a branch to be bound to parent, particularly if you have a treeless mainline on a server?

- Scott 

P.S. Vincent sorry for double sending...forgot reply-all