AW: Newbie question--how to clone hg itself?
Matt Mackall
mpm at selenic.com
Fri Oct 31 15:59:00 UTC 2008
On Fri, 2008-10-31 at 09:24 +0100, Kauker, Hubert wrote:
> > > What is the equivalent for grabbing the latest hg release?
> >
> > hg clone http://selenic.com/repo/hg-stable
>
> I tried the following alternatives to compare them.
> Is there any explanation for the vast timing differences?
Yes, you're pulling from a fairly busy server sitting on the slow end of
a DSL pipe (< 1Mbps). The server runs multiple virtual machines, all of
which have substantial mail, web, and shell user load. The virtual
machine in question generally reports a load average > 2, where 1=no
idle CPU.
Here's the result of 10 clones (with -U) over local wireless to the same
machine:
Time: real 13.270 secs (user 8.280+0.000 sys 1.390+0.000)
Time: real 10.890 secs (user 8.020+0.000 sys 1.520+0.000)
Time: real 12.510 secs (user 8.260+0.000 sys 1.370+0.000)
Time: real 17.660 secs (user 8.110+0.000 sys 1.480+0.000)
Time: real 16.310 secs (user 8.290+0.000 sys 1.480+0.000)
Time: real 11.440 secs (user 7.960+0.000 sys 1.530+0.000)
Time: real 25.230 secs (user 8.040+0.000 sys 1.430+0.000)
Time: real 10.850 secs (user 8.030+0.000 sys 1.520+0.000)
Time: real 11.270 secs (user 8.000+0.000 sys 1.470+0.000)
Time: real 12.040 secs (user 8.090+0.000 sys 1.430+0.000)
Similar 'user' and 'sys' time to your machine, totalling about 9.5
seconds. Seems to hit a floor at about 10.8 seconds, which I suspect is
as fast as the server can go (and is only slightly slower than the
client can receive).
It's also worth noting that network latency shouldn't be an issue here
as the bulk of the data is sent as a single HTTP response.
In other words, your problem is entirely one of available bandwidth.
You should try a mirror:
http://hg.intevation.org/mercurial/
--
Mathematics is the supreme nostalgia of our time.
More information about the Mercurial
mailing list