Default performance of 'lp:foo' URLs

[Michael Hudson]

Stephen J. Turnbull wrote:
> Gordon Tyler writes:
> 
>  > SSH handshaking when all you want to do is pull the latest couple of
>  > revisions into your clone of trunk is somewhat annoying.
> 
> Huh?  You can observe the difference between 5ms and 15 ms?  I can't.

Um, how do you manage that?

mwh at grond:trunk$ time ssh python.net true

real	0m5.752s
user	0m0.010s
sys	0m0.000s

I am in New Zealand, the server is in Germany, about 380ms ping.  If I
use connection multiplexing, it is much faster, but it would have a hard
time being faster than, well, 380 ms :-)  Heck, 15ms isn't long enough
for *light* to get from me to the server, in a straight line!

> Maybe you're seeing something very different, but I use ssh URLs with
> Mercurial all the time and just don't notice any difference between
> them and http URLs -- even when I was looking, it's just way below my
> JND.  Most of the time is spent updating the tree.  I'm outside of
> Tokyo, the server (alioth.debian.org) seems to be in Amsterdam
> according to whois.  I do it with git all the time to a server on the
> same local network.  Same result.  I've never heard any complaints
> from my colleagues about delays with SSH, either.

I would love to know how that is managed.

>  (On the other hand,
> they're all still euphoric -- after 18 months :-) -- about the sub-
> second diff capability.)

Cheers,
mwh