[SciPy-User] Test failures with SciPy-0.10.1 and Mac OS X Lion
Mon Mar 5 09:30:22 CST 2012
On Mon, Mar 5, 2012 at 06:35, Pauli Virtanen <email@example.com> wrote:
> The failures seem to indicate that (single-precision) linear algebra
> routines are not functioning correctly. These are probably a sign of
> some Veclib / Fortran ABI mismatch related to Fortran functions --
> similar problems have been seen e.g. when mixing g77 and gfortran
> compilers (also on other platforms than OSX), and on OSX adding Veclib
> to the mix seems make the situation even more finicky.
> Could you file a ticket:
> Currently, we use the Veclib's C interface for complex-valued functions,
> but possibly we should do this for all Fortran functions to work around
> these problems.
In my original email I said that the compilers from Xcode-4.3 had been
used, this is not the case. Everything should have been build using
the compilers from gcc-4.4.6. Xcode-4.3 is installed on the machine
but it's compilers aren't used for building scipy itself.
More information about the SciPy-User