Wednesday, September 21, 2005

LSB comments

Ulrich's rant about LSB has a number of issues where he just wrote wrong information. Since he's mentioning some of these for SUSE Linux 10.0, let me point out the following:

  • The filing speaks about minimal requirements, e.g. "Minimum Processor Speed". We did not test this on such old hardware but I'm glad that we support such old hardware. The tests were actually done on a new Athlon64 3000+ single processor system (running a 32-bit kernel).
  • Our engineers looked already during the development phase of LSB 3.0 at the test versions of the certification testsuite and reported bugs in the testsuite before LSB 3.0 was released. During our final application we nevertheless asked for new waivers and also could reuse some of the already filled waivers by others. Note that waivers are anonymous in the database and not shown anywhere. So Ulrich's claim "without the people reporting any problems and requesting waiving the test" is based on his unability to find company names in an anonymous database - and on the way that the LSB does not do a full disclosure of the reports filled by the companies.

I agree with Ulrich that the testsuite is not in the best shape, this is something that hit us as well. We only filled the ia32 LSB 3.0, but tested others archs during the 3.0 pre-testing as well. Ulrich seems to have to support all the Red Hat architectures and since the testsuite seems to have been originally developed on ia32 systems, shows less problems there than on other platforms.

I'm glad that we were the first to have a LSB 3.0 certification - and I wonder whether Ulrich considers this a "lost battle" and therefore started his rants against LSB and us.

1 comment:

Amrith said...

I am using the LSB test cases for a different platform to see the compatibility results. I am confused to see the Actual tests are written in Fortran. What could the reason for this