Hi, I just detected a problem that *looks like* a serious problem with either Python or libxml2 (or something else?).
I have a Python program that operates on data files and that reads an XML configuration file. It does some work for every data file. In the code, there are some assertions which I inserted in the past (luckily), related to presence of some XML attributes etc. On RHEL5 beta2 this program fails with that assertion on random data files, but only on RHEL5, not on RHEL4, while using exactly the same data files and configuration files. Some observations: - Sometimes it goes wrong with the 10th data file, sometimes with the 15th, sometimes with the first. - In all those cases the same assertion fails. - In some cases the same failing data file stays failing the next run, over and over, but the failure is gone the first run after removal of the .pyc files. However, also without .pyc files and having a non-writeable directory, the error at some points occurs, so it's not purely related to .pyc files. - I tried it on multiple physical systems, all with the same error occurring at random data files. - I tried it on i386 and x86_64 OS's, both have with the same errors (but this is in all cases the same physical Opteron system). - I did the RHEL4 run (processing 1000+ data files without any error) on the same physical machine and filesystem as the RHEL5 tests. IMHO, this is a non-determistic behaviour for deterministic data. Unfortunately, I can't easily isolate the problem and post that, although I might try to do that later. Are there any known problems that could relate to what I describe or are there any suggestions for things to try to find the cause of this weird behaviour? Cheers, -- -- Jos Vos <[EMAIL PROTECTED]> -- X/OS Experts in Open Systems BV | Phone: +31 20 6938364 -- Amsterdam, The Netherlands | Fax: +31 20 6948204 _______________________________________________ rhelv5-beta-list mailing list rhelv5-beta-list@redhat.com https://www.redhat.com/mailman/listinfo/rhelv5-beta-list
