Fastest way to find repository heads?

[Robert Collins]

On Thu, 2009-07-09 at 00:25 +0200, Jelmer Vernooij wrote: 
> Robert Collins wrote:
> > On Thu, 2009-07-09 at 00:04 +0200, Jelmer Vernooij wrote:
> >   
> >> Admittedly this is something that should ultimately be fixed by making
> >> "bzr branch" do the right thing and leave a working branch behind, but
> >> until then I think bzr-git should use dead and live heads rather than
> >> just live heads. 
> >>     
> >
> > While I think it would be nice to do that, I think in terms of tradeoffs
> > its not worth doing that. Last time I checked, git doesn't do that, and
> > neither does hg.
> >   
> What tradeoffs exactly? As a user, I'd be happy to pay a small
> performance overhead to have "bzr branch" be resumable, it'd be a large
> usability improvement.

Depending on how frequently you want checkpoints, some multiple of
bandwidth to clone, with corresponding CPU and memory overhead. We
currently have one checkpoint - the whole tree. In trees where every
object has 2 groups because of size, adding a single checkpoint would
have minimal overhead. If the tree had 1 group for most objects (so
things are highly compressed) adding a single checkpoint would double
the transfer overhead. Adding two checkpoints would triple the overhead.

-Rob 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20090709/a299ca92/attachment.pgp