[IPython-user] Problem doing svn co behind proxy

Fernando Perez fperez.net@gmail....
Mon Aug 13 04:59:31 CDT 2007


On 8/13/07, David Cournapeau <david@ar.media.kyoto-u.ac.jp> wrote:

> I don't know the size of envisage, but I know that bzr with numpy and
> scipy, with the full history, is faster to get log and so on than svn
> (this is of course due to network latency, but concretely, those are
> useful advantages of DVCS). I tried both hg and bzr on big projects to
> get an idea (kernel size), and I found both of them unusable for those
> tree size (i.e. knowing which files changed takes more than 1-2 seconds,
> commit are really slow, etc...).
>
> Anyway, this is all moot without "hard numbers".  V. Vaino suggested a
> script to use hg, I can try to do something for bzr on my side, and then
> we can compare them and you can make the decision according to those
> results . Would this be considered useful ? (again, I do not care about
> bzr vs hg; I am more familiar with bzr, but I believe the advantages of
> DVCS, specially for projects around scipy, to largely outweight any
> difference between bzr and hg)

I'm inclined to go with hg because:

- As Ville mentioned, at least on the net there seems to be a lot of
evidence of the speed difference between hg and bzr being huge.

- In my (admittedly quick and dirty, non-scientific, non-reproducible)
tests, bzr felt painfully slow.

- SAGE tried both and quickly went for hg for speed reasons, managing
their tree with bzr apparently was just a non-starter.

- When Enthought recently discussed the possibility of moving to a
DVCS, hg was their choice.  This is an area where I'd rather use the
same tool that other related projects (SAGE, Enthought tools) are
using.

- The scipy.org server already has hg but not bzr.  One less thing for
me to install.

The one thing I like about bzr is that *any* http-accessible tree is
valid for cloning, while hg needs a server process to run.  I don't
understand why hg has this requirement, it seems like a great idea to
follow bzr's lead here.

We'll see where we end up, I'll talk to the Enthought crowd this week
and ask for their advice.

Cheers,

f


More information about the IPython-user mailing list