CESA-2005:685 Low CentOS 4 s390(x) mysql - security update
Posted on: 10/06/2005 05:47 PM
CentOS Errata and Security Advisory 2005:685https://rhn.redhat.com/errata/RHSA-2005-685.html
The following updated files have been uploaded and are currently syncing to the mirrors:
There is feature in hercules, which makes it refelect sometimes floating point test errors for aaplications due different accuracy of real iron floating point and the host emulating it. It's kind of must have design feature where hercules does pass the floating point operations to host fp unit directly. It would slow down ten fold or more if it would not do so.
This efectively means that for some packages the test phase fails under hercules even tho it's not failing on real iron. I've had access to real iron (not currently) and been able to verify that very same compilers + other parts makes testing work on real iron and fail under hercules. Resulting binaries still being good on oth platforms.
Therefore i just have to trust that binaris are ok even tho the internal test procedures are failing. This is the case for s390 with mysql. s390x compiles + tests just fine. I start to make a note in these announcements if i have been forced to skip the package internal testing phase for being able to make binary RPMS. Something like there is below for s390.
It's just in case somene should rely on this distribution on real iron as i am not able to verify/support this scenario. With hercules we're all in same wagon - if it's not working, it's fixable for us
s390: (had to disable check/test)