Justin Erenkrantz wrote:
> On Nov 3, 2007 11:28 AM, Philip Martin <philip@codematters.co.uk> wrote:
>> Philip Martin <philip@codematters.co.uk> writes:
>>
>>> I think your machine should be able to run the tests much faster, my
>> That's much faster than your machine currently runs them, not much
>> faster than my machine.
>
> Aha. I finally found the real culprit. =)
>
> Karl compiled svn with --enable-maintainer-mode and I believe Philip
> did too. This defines SVN_DEBUG - which causes
> svn_sleep_for_timestamps to now be conditional on the envvar
> SVN_SLEEP_FOR_TIMESTAMPS. The test suite defines this to be 'no'. I
> compiled in 'release' mode (so SVN_DEBUG was not defined) - so
> removing the sleep causes me to run basic_tests.py in 10 seconds on my
> box.
I've always built with --enable-maintainer-mode, and for me there is a real
difference between using /dev/random and /dev/urandom. My entropy was
clearly getting sapped using /dev/random. I could run the test suite, watch
it operate for a while before slowing to a crawl, and any time I'd jiggle my
mouse, I'd get a bit of speed-up. Just running 'make check --cleanup'
(ra-local) -- even on a ramdisk -- used to take me nearly an hour. Since
the switch to urandom, I can't even heat and eat a plate of leftover
spaghetti before it completes.
The sleep-for-timestamps thing may be somewhat interesting, but I don't
think it is necessarily the "real culprit".
--
C. Michael Pilato <cmpilato@collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand
Received on Mon Nov 5 14:31:44 2007