[RFC] make checkouts as much like bound branches as possible

[Olaf Conradi]

On 28/03/06, Aaron Bentley <aaron.bentley at utoronto.ca> wrote:
> Olaf Conradi wrote:
> > On 28/03/06, Aaron Bentley <aaron.bentley at utoronto.ca> wrote:
> >
> >>Olaf Conradi wrote:
>
> >>>A lightweight checkout with a bound branch has the benefit of
> >>>unbinding and taking your laptop offline, as the lightweight checkouts
> >>>were part of a local shared repository.
> >>
> >>So this is:
> >>A bound branch at ~/bound-branch
> >>A lightweight checkout of ~/bound-branch at ~/checkout
> >
> > No, the lightweight checkout is the same as the bound branch
> >
> > ~/checkout $ bzr bind http://blabla
>
> This command will cause the lightweight checkout's branch to be bound.
> The branch of a lightweight checkout is never at the same location as
> the checkout, by definition.  Where is it?

Ah, that is what I misunderstood.

I thought the double binding already worked for lightweight checkouts..
I assumed the new change would go to both the new bound location and
the shared repository of the checkout.

Just tested it by binding a lightweight checkout and committing, the
update only appeared in the bound branch, not repo. I tested the
binding and commiting before, but somehow missed that it did not
update the repo.

> > I am saying that any type of branch should have the option to bind
> > against some other tree. Just as it has the option to specify a parent
> > and push location.
>
> Okay.  This implies that it should be possible to bind a branch that is
> the master of a bound branch, i.e. double-binding.
>
> > That the only difference between a lightweight checkout and a normal
> > checkout (or bound branch - whatever one prefers) is the location of
> > the revisions. In the same .bzrdir or in some shared repository.
>
> Okay, so you're in favour of my proposal then?

Jup :)

Cheers
 -Olaf