On Mon, Jul 19, 2004 at 04:45:41AM +0000, Clement Vasseur wrote:
Clement Vasseur <nitro(a)epita.fr> wrote
(times in seconds, on sandrock)
953.18 ./compile/920428-2
859.20 ./compile/20020304-1
822.01 ./execute/20020108-1
[...]
0.04 ./compat2/scalar-by-value-1_x
0.04 ./compat2/scalar-by-value-1_main
0.04 ./compat2/fnptr-by-value-1_y
Hum, I just realized that these timings are plain wrong. On Sandrock
there is not enough space in /tmp for all the temporary files for the
whole testsuite, so from a certain test, every tests failed immediately,
that's why there are timings as low as 0.04 second.
So, there should be a lot more time-consuming tests than the ones I
show here.
I will remake the test timings on hercule later today.
This temporary files problem is very annoying. Can anyone take care of
it? (I will put a task on gforge as soon as I come back from school).
I have written a little shell script to solve this problem:
#! /bin/sh
# -*- sh -*-
if [ $# -eq 0 ]; then
echo "usage: clean-str command..."
exit 1
fi
eval "$@"
status=$?
rm /tmp/StrategoXT*
exit $status
This script is available at ~despre_n/bin/clean-str
It is probably not the best solution but it works fine.
--
Nicolas Despres
nicolas.despres(a)lrde.epita.fr