Thousands of branches...

[Giorgos Keramidas]

On Thu, 24 Jul 2008 11:22:51 -0400, Doug Philips <dgou at mac.com> wrote:
> [snip interesting test with 1000 local branch names]
>
> After having set up this artificial repo, the 'hg branches' command on
> both machines is still I/O bound to the screen.
>
> hg log (so far as I've tested) does not seem to be slowed down either.

Hi Doug,

That's interesting, and pretty much what I'd expect from skiming through
the source of "hg branches" and "hg log" :-)

> Net of the net is that it seems that giving branches names, even for
> lots and lots of short-ish branches won't drag us down for quite a few
> years, if at all.

Cool!  Thanks for sharing the results.

If you have kept a copy of the file `branch.names' or of the test
repository, feel free to email me a bundle of all changesets, created
with:

    hg bundle --base null /var/tmp/branchtest.hg

It would be interesting to see what timing results "log" and "branches"
give on my laptop (running FreeBSD and the Linux workstation I have at
home).