Package: octave2.9 Severity: important Dear Octave maintainers,
First I must apologize for reporting a bug I can't reproduce. But if this is a real bug, it's rather critical (I would have reported it as RC if I could reproduce), and it may be architecture dependent, so here it goes: A user reported to Ubuntu that octave 2.9.9-6 (the source package is unchanged, just rebuilt for Ubuntu) generates wrong results for complex matrix multiplication [1]. The discussion on upstream mailing list [2] confirmed that it's reproducible on Debian as well with 2.9.12 [3]. Both cases are on amd64 system. 1. https://bugs.launchpad.net/ubuntu/+source/octave2.9/+bug/117517 2. http://www.cae.wisc.edu/pipermail/bug-octave/2007-May/thread.html#2494 3. http://www.cae.wisc.edu/pipermail/bug-octave/2007-May/002498.html The conclusion on upstream mailing list seems to be that Debian package is missing the -ff2c flag when using gfortran to compile octave and linking to g77-compiled BLAS/ATLAS libraries. As I understand neither octave or Fortran ABI, I am just forwarding this bug to Debian to raise your attention. If a Debian maintainer can test on an amd64 system, it will be greatly appreciated. Ming 2007.06.02 -- octave is linking incorrectly BLAS/ATLAS libraries https://bugs.launchpad.net/bugs/117517 You received this bug notification because you are a member of Ubuntu Bugs, which is the bug contact for Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs