I've mentioned in the past that our relational database product has ceased
working with 4.0F. Our engineers have been working on this non-stop for a
month now and are closing in on a reproducer program that eliminates any
part of our product to demonstrate the problem. To make a long story short
what they're finding is that the UNIX sleep() command appears to
occasionally corrupt a floating point register, which eventually causes
either a floating point exception or segmentation fault. The problem seems
to occur under heavy load (LOTS of multi-threading going on).
I mention this for 2 reasons:
1. Has anyone heard of anything similar?
2. An FYI in case you see something similar.
BTW, we're a reseller, formerly an ASAP member and now a CSA member, and
supposedly a close partner (our product only runs on Alpha) but all our
access numbers are coming up invalid (expired?) trying to get through to
the support center and really have no other technical contacts into Compaq
other than this list. I mention this to vent some frustration but also in
answer to anyone wondering if we've escalated this to support. The answer
is no, we haven't, because we don't seem to be able to. At this point I'm
simply trying to buy a support contract, no easy trick in and of itself.
Chris
Received on Fri Sep 10 1999 - 15:37:30 NZST