Comment 9 for bug 332002

2009/3/16 Gábor Melis <email address hidden>:
> I committed the stuff referenced above as 1.0.26.4 - 1.0.26.6. In all
> likelihood, it won't solve your problems, but it would still be
> interesting to see how the test added in "less pessimal waitqueues" does
> on your system.

I recently tried the semaphore-test you emailed to me again on
sbcl 1.0.25.56, which was the release I initially complained about
as being broken for me. Surprisingly, although my workstation had
not been rebooted since the last runs and the sbcl binary had not
changed, I had to run the test about 10 times before I saw a
failure. Previously, about half the runs would fail for me.

Anyway, I built sbcl 1.0.26.8 and tried the test with that version.
It seems to work reliably now and generally the test takes a lot
less time to run than it used to take with 1.0.25.56. I have not yet
seen it fail.

bob