Re: [f2py] integers = 2226756?
"Michael Colonno" <mcolonno-FGKo4X94FMn2fBVCVOL8/[email protected]> Thu, 24 Mar 2011 12:58:15 -0700
| Newsgroups | gmane.comp.python.f2py.user |
|---|---|
| Message-ID | <[email protected]> |
Tried that, same result unfortunately.
Thanks,
~Mike C.
From: f2py-users-bounces-Y4l6ocDipWCuvFJfX82//[email protected] [mailto:f2py-users-bounces-Y4l6ocDipWCuvFJfX82//[email protected]]
On Behalf Of [email protected]
Sent: Thursday, March 24, 2011 12:48 PM
To: For users of the f2py program
Subject: Re: [f2py] integers = 2226756?
Try
integer*8, intent(in) :: N
On , Michael Colonno <mcolonno-FGKo4X94FMn2fBVCVOL8/[email protected]> wrote:
> Thanks for the help and strange indeed. After some more
research this appears to be platform dependent. I'm using f2py on Win 7
64-bit (which seems to be its least favorite platform) with the Intel
compiler suite (12.0) and VS 2010. Using the same compilers on the same
system in Linux (CentOS 5.5) the test module works fine, integers are passed
with the correct value. I did have to do a bit of hacking in the
mscv9compiler.py and intel.py files to get the flags to match up for proper
linking, but none of the flags I tweaked should, as far as I can tell,
impact the passing of basic data types (almost all linker flags). I did
switch the !f2py directive in and out with no effect. My first thought was a
32-bit / 64-bit setting somewhere in the maze of dependencies between VS and
the Intel compilers on 64-bit platforms but so far no luck. I will run the
test suggested and report results. Thanks, ~Mike C.
From: f2py-users-bounces-Y4l6ocDipWCuvFJfX82//[email protected] [mailto:f2py-users-bounces-Y4l6ocDipWCuvFJfX82//[email protected]]
On Behalf Of [email protected]
> Sent: Wednesday, March 23, 2011 11:26 PM
> To: For users of the f2py program
> Subject: Re: [f2py] integers = 2226756? This is strange. I'm not able to
reproduce it with gfortran=4:4.5.2-2 python-numpy=1:1.4.1-5 in Debian. It
can put in int8-int64 without problems. What compiler are you using? What
version of f2py? It might be worthwhile looking at the automatically
generated C code (what you get running f2py without -c) to see what's
happening.
> The "f2py_rout_test_test" in particular would be of interest.
>
> Kevin
>
> On , Michael Colonno mcolonno-FGKo4X94FMn2fBVCVOL8/[email protected]> wrote:
> >
> > Hi ~
> >
> > I have what I hope is a very simple issue with integers. My
extremely simple F90 code:
> >
> > subroutine test(N)
> >
> > implicit none
> >
> > !f2py integer, intent(in) :: N
> > integer :: N
> >
> > print *, "N = ", N
> >
> > end subroutine
> >
> > I compile this with F2PY and get in Python:
> >
> > >>> import test
> > >>> test.test(13)
> > N = 2226756
> > >>> N = 7
> > >>> test.test(N)
> > N = 2226756
> >
> > I get the same value for any value I input. If I change the argument N
to a real, I get the correct value. Is there some kind of incompatibility
re: integers from Python? Please let me know if I'm going crazy.
> >
> > Thanks,
> > ~Mike C.
> >
> >
>
>
_______________________________________________
f2py-users mailing list
f2py-users-Y4l6ocDipWCuvFJfX82//[email protected]
http://cens.ioc.ee/mailman/listinfo/f2py-users