[Scipy-tickets] [SciPy] #1618: Testsuite failures with 0.10.1 under Mac OS X Lion
SciPy Trac
scipy-tickets@scipy....
Wed Aug 1 09:12:52 CDT 2012
#1618: Testsuite failures with 0.10.1 under Mac OS X Lion
----------------------+-----------------------------------------------------
Reporter: ramercer | Owner: somebody
Type: defect | Status: new
Priority: normal | Milestone: Unscheduled
Component: Other | Version: 0.10.0
Keywords: |
----------------------+-----------------------------------------------------
Comment(by mike.wimmer):
Replying to [comment:4 pv]:
> The correct solution is probably to not use any Fortran FUNCTIONs
(SUBROUTINEs should be OK, though), and instead use the C ABI, as we do
with the complex-valued cases. These Fortran ABI issues in Veclib might
possibly be worked around with some magic compiler options, but it seems
best to just avoid the mess.
As a short term solution I would like to note that at least on MacOSX
10.8, specifying the "-ff2c" flag does solve the problem. As noted in the
macresearch.org thread you cite above, the Accelerate framework adheres to
the g77 conventions, and -ff2c does force gfortran to do exactly that.
In principle this is hence the correct way to compile.
It appears that someone in the above thread mentioned had problems
together with 64 bit code, which people later in the thread identified as
an Accelerate bug. Now at least for me, I have never seen this particular
(-ff2c together with -m64) bug, and it might very well have been fixed by
Apple. This would make the "-ff2c"-switch the correct solution in any
case.
--
Ticket URL: <http://projects.scipy.org/scipy/ticket/1618#comment:5>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.
More information about the Scipy-tickets
mailing list