Re: Testing the new VC code

[Eric S. Raymond]

Michael Albinus <address@hidden>:
> 1. Test: Check a file not under vc
>
> Your version is even a little bit faster. IIRC, the heuristic functions
> weren't such good for Tramp, because they have used process calls
> instead of file existence checks. Tramp internal optimizations do not
> work then.

Yes, that is one reason I was expecting the change to make little difference

> 2. Test: Check a file under CVS control. The CVS repository is on
>    savannah
> 
> Your version is not bad, but a factor of 13 slower. So if you have a
> slow connection to your CVS repository, caching would help.

This is the only case turned up in your testing that concerns me.  It
may be an argument for restoring some of the state-heuristic machinery
in the CVS back end only. Or maybe not - because of your git results 
I'd need a bit more persuading.  See below.

> 4. Test: Check a file under Git control
> 
> Again, your version is slower (15%). More surprising, both versions are
> much slower than with Bzr. I guess one could improve the code for git.

This suggests that there is significant noise in your profiling, because
the git back end had no local caching to begin with.  There must be some
external source of variation.

I suspect that your test is quite sensitive to short-term fluctuations
in network latency and a lot of what you were measuring was actually
that. Otherwise the checkin running a bit *faster* would be hard to
explain.

In any case, none of the differences seem worth getting excited about.
I'll keep an eye on CVS latency, but I won't reintroduce complexity
against it unless we get complaints from real users.  With response
times of a quarter second I think that is unlikely - it's not that far 
above the minimum ergonomic threshold of 0.17sec below which humans
simply cannot notice latency at all.

(A spinal reflex arc is about 0.10 seconds.  Human nerve conduction velocity -
the "speed of thought" - is not actually very high.)
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>