1.13 vs 1.14rc1 performance summary

[Andrew Bennetts]

Ian Clatworthy wrote:
> John Arbash Meinel wrote:
[...]
> > Bundle:send	23.9	1.2	7.1	2.6 bzr send -o zzzBundle.path
> > 
> > 23.9s => 1.2.. I know there were a couple of improvements here,do you
> > think these are genuine? I guess emacs has a *very* long mainline, so it
> > might be possible. I don't think emacs has a much larger number of commit
> 
> I hope so, given it was me who tuned send during 1.14. :-) Emacs has
> 105K revisions and I changed some things from being O(N) to not being that.
> My _iter_just_entries() change is helping as well - it was added to make
> send faster in brisbane-core and landed indirectly through there.

Also, I changed write_revisions in bzrlib.bundle.serializer.v4 to get all
the revisions in a single get_revisions call, rather than one at a time.
I'd expect that to make a large difference too (assuming that benchmark is
sending multiple revisions).

So if you tuned send as well, then I can easily believe it has improved
dramatically!

-Andrew.