Thoughts on push performance

[Andrew Bennetts]

John Arbash Meinel wrote:
> Andrew Bennetts wrote:
[...]
> | Can you see any reason why we shouldn't just do this to _basic_push:
[...]
>
> I'm guessing I put together something for 'update_revisions' because of
> something I was seeing with 'bzr pull', and the push side never got the same
> treatment.
>
> Looking at 'pull' we have a very nice:
>
> self.update_revisions(source, stop_revision, overwrite=overwrite)
>
> With no try/except Diverged, etc. So +1 on your fix.

Hah.  I just experienced <https://bugs.launchpad.net/bzr/+bug/234229>: 
"bzr push -r NNN --overwrite" is broken at the moment.  The change I proposed
above fixes it.  I'm working on making a proper patch with test at the moment.

So it's not only simpler, it's less buggy.  I guess I shouldn't be surprised
that these two things go together!

-Andrew.