New discovery code

[Peter Arrenbrecht]

On Wed, May 4, 2011 at 5:29 AM, Matt Mackall <mpm at selenic.com> wrote:
> Peter's new discovery code from the sprint is in the tree and I've got
> it running on http://selenic.com/hg and /repo.
>
> a) It's really much faster. I've got a test here of a kernel repo that's
> over 10k csets behind tip. Old discover takes at least a few dozen round
> trips and about a minute, new discovery takes 2 round trips and about a
> second.

BTW, folks interested in how much traffic this could eliminate on
their servers should compute the ratio of "heads" vs
"branches"+"between" requests (try to eliminate the calls to between
that serve as a hello, though). I think this should give a rough
indication of the chattiness of current tree discovery. New set
discovery should be in the range of 2-3 calls to "known" per one
"heads".

-parren