Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread Carl, Andrew F (AS)

Then how about known combinations of version numbers for g77/gcc,
python, and numpy? The gist would be to convey something that is known
too work together. Some of the magic under the hood is looking at
system environmental variables, and the Intel fortran compiler flags
wreak havoc, requiring the uninstall of the intel compiler to prevent
corruption of the gnu95 setup w/ gfortran. Something on where to look to
see whats going on would be helpful!

Thanks,
Andy

-Original Message-
From: numpy-discussion-boun...@scipy.org
[mailto:numpy-discussion-boun...@scipy.org] On Behalf Of David
Cournapeau
Sent: Tuesday, June 09, 2009 7:43 PM
To: Discussion of Numerical Python
Subject: Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

Carl, Andrew F (AS) wrote:

 Would it be a reasonable request, that the F2PY Windows web page 
 contain known combinations of version numbers for Python, Numpy and 
 Gfortran verified to play nice? Some references as to queried compiler

 system environmental variables would be useful also.

I have added some numpy.distutils support for gfortran on windows, but
Windows + gfortran  + numpy is unlikely to work well unless you build
numpy by yourself with gfortran. I have actually been considering a move
to gfortran for windows builds, but I would prefer waiting mingw to
officially support for gcc 4.* series.

cheers,

David
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread David Cournapeau
On Thu, Jun 11, 2009 at 12:30 AM, Carl, Andrew F (AS)a.c...@ngc.com wrote:

 Then how about known combinations of version numbers for g77/gcc,
 python, and numpy?

Any of them should work. Numpy on windows is built with g77/gcc as
available from the official MinGW installer.

. Something on where to look to
 see whats going on would be helpful!

Something like python setup.py build_ext --fcompiler=gnu

should force to look for g77 instead of the Intel fortran compiler.

David
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread Carl, Andrew F (AS)
The default finds both the g77  gfortran compilers. The issue is that
the flags associated w/ both appear to be corrupted by the contents of
the intel fortran environmental variables (i.e. content matched, removal
of the intel system environmental variables made the problem go away),
resulting in the example problem failing. Used f2py/diagnose.py to
verify. It's just that it took about six python install/uninstall
cycles. It would be nice to enable others to avoid the problem via a
heads-up on the F2Py Windows web page.


-Original Message-
From: numpy-discussion-boun...@scipy.org
[mailto:numpy-discussion-boun...@scipy.org] On Behalf Of David
Cournapeau
Sent: Wednesday, June 10, 2009 8:44 AM
To: Discussion of Numerical Python
Subject: Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

On Thu, Jun 11, 2009 at 12:30 AM, Carl, Andrew F (AS)a.c...@ngc.com
wrote:

 Then how about known combinations of version numbers for g77/gcc, 
 python, and numpy?

Any of them should work. Numpy on windows is built with g77/gcc as
available from the official MinGW installer.

. Something on where to look to
 see whats going on would be helpful!

Something like python setup.py build_ext --fcompiler=gnu

should force to look for g77 instead of the Intel fortran compiler.

David
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread Andrew Hawryluk
 -Original Message-
 From: numpy-discussion-boun...@scipy.org [mailto:numpy-discussion-
 boun...@scipy.org] On Behalf Of David Cournapeau
 Sent: 9 Jun 2009 8:43 PM
 To: Discussion of Numerical Python
 Subject: Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content
 
 Carl, Andrew F (AS) wrote:
 
  Would it be a reasonable request, that the F2PY Windows web page
  contain known combinations of version numbers for Python, Numpy and
  Gfortran verified to play nice? Some references as to queried
 compiler
  system environmental variables would be useful also.
 
 I have added some numpy.distutils support for gfortran on windows, but
 Windows + gfortran  + numpy is unlikely to work well unless you build
 numpy by yourself with gfortran. I have actually been considering a
 move to gfortran for windows builds, but I would prefer waiting mingw
 to officially support for gcc 4.* series.

I was able to get gfortran working on Windows just a few weeks ago. The
only problem I had was that I needed Python = 2.5.3 before it would
work. See issue #2234 in
http://www.python.org/download/releases/2.5.4/NEWS.txt

Andrew
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread David Cournapeau
On Thu, Jun 11, 2009 at 1:22 AM, Andrew Hawrylukhawr...@novachem.com wrote:

 I was able to get gfortran working on Windows just a few weeks ago. The
 only problem I had was that I needed Python = 2.5.3 before it would
 work. See issue #2234 in
 http://www.python.org/download/releases/2.5.4/NEWS.txt

That's not really the problem I had in mind - rather, if you compile
your extension with gfortran, since numpy is built with g77 on
windows, you can get into hairy situations since g77 and gfortran are
not compatible. If you build numpy yourself with gfortran (which
should work), then there is no problem.

David
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread Carl, Andrew F (AS)

Got gfortran (dated 2009-04-21) working w/ python 2.5.2, numpy 1.3.0 
MinGW 5.1.4, and after removing the intel fortran system environmental
variables (plus six python installs, a intel fortran uninstall, and a
day and a half of head scratching).

Andy

-Original Message-
From: numpy-discussion-boun...@scipy.org
[mailto:numpy-discussion-boun...@scipy.org] On Behalf Of Andrew Hawryluk
Sent: Wednesday, June 10, 2009 9:22 AM
To: Discussion of Numerical Python
Subject: Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

 -Original Message-
 From: numpy-discussion-boun...@scipy.org [mailto:numpy-discussion- 
 boun...@scipy.org] On Behalf Of David Cournapeau
 Sent: 9 Jun 2009 8:43 PM
 To: Discussion of Numerical Python
 Subject: Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content
 
 Carl, Andrew F (AS) wrote:
 
  Would it be a reasonable request, that the F2PY Windows web page 
  contain known combinations of version numbers for Python, Numpy and 
  Gfortran verified to play nice? Some references as to queried
 compiler
  system environmental variables would be useful also.
 
 I have added some numpy.distutils support for gfortran on windows, but

 Windows + gfortran  + numpy is unlikely to work well unless you build 
 numpy by yourself with gfortran. I have actually been considering a 
 move to gfortran for windows builds, but I would prefer waiting mingw 
 to officially support for gcc 4.* series.

I was able to get gfortran working on Windows just a few weeks ago. The
only problem I had was that I needed Python = 2.5.3 before it would
work. See issue #2234 in
http://www.python.org/download/releases/2.5.4/NEWS.txt

Andrew
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread Bruce Southey
On 06/10/2009 11:41 AM, Carl, Andrew F (AS) wrote:
 Got gfortran (dated 2009-04-21) working w/ python 2.5.2, numpy 1.3.0
 MinGW 5.1.4, and after removing the intel fortran system environmental
 variables (plus six python installs, a intel fortran uninstall, and a
 day and a half of head scratching).

 Andy

Hi,
What is the operating system (version and type) you are using?
Also, could you please list the steps that worked or an overview of them 
as it may help other people?

Thanks
Bruce
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-10 Thread Carl, Andrew F (AS)

PC w/ Windows XP [Version 5.1.2600] 
Steps:
1) Follow steps 1 thru 3 from F2py Windows web page, in my case as
follows:
a) ActiveState ActivePython 2.5.2.2
b) Numpy 1.3.0
c) gFortran (dated 2009-04-21)
2) Install MinGW 5.1.4 (g++, g77, make)
3) Check system environmental variables as follows:
a) PATH:
C:\gfortran\libexec\gcc\i586-pc-mingw32\4.5.0;C:\gfortran\bin;C:\MinGW\b
in
b) C_INCLUDE_PATH: C:\gfortran\include
4) Run C:\Python25\Lib\site-packages\numpy\f2py\diagnose.py and review
output for g77  g95 fcompilers found  contents of flags:
a) GnuFCompiler instance: c:\MinGW\gin\g77.exe
b) Gnu95FCompiler instance: c:\gfortran\bin\gfortran.exe
c) If you don't see items a  b above, don't bother trying
the example problem from the F2py Windows web page
5) same as step 5 from F2py Windows web page

To make matters worse, an install/uninstall of ipython from Add
or Remove Programs broke it, requiring the entire process to be
repeated (i.e. uninstall/re-install numpy  scipy alone did not work).
THIS IS SCARY (and frustrating)!

Andy

-Original Message-
From: numpy-discussion-boun...@scipy.org
[mailto:numpy-discussion-boun...@scipy.org] On Behalf Of Bruce Southey
Sent: Wednesday, June 10, 2009 11:33 AM
To: Discussion of Numerical Python
Subject: Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

On 06/10/2009 11:41 AM, Carl, Andrew F (AS) wrote:
 Got gfortran (dated 2009-04-21) working w/ python 2.5.2, numpy 1.3.0 
 MinGW 5.1.4, and after removing the intel fortran system environmental

 variables (plus six python installs, a intel fortran uninstall, and a 
 day and a half of head scratching).

 Andy

Hi,
What is the operating system (version and type) you are using?
Also, could you please list the steps that worked or an overview of them
as it may help other people?

Thanks
Bruce
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


[Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-09 Thread Carl, Andrew F (AS)
Would it be a reasonable request, that the F2PY Windows web page
contain known combinations of version numbers for Python, Numpy and
Gfortran verified to play nice? Some references as to queried compiler
system environmental variables would be useful also.
Thanks,
Andy
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion


Re: [Numpy-discussion] Inquiry Regarding F2PY Windows Content

2009-06-09 Thread David Cournapeau
Carl, Andrew F (AS) wrote:

 Would it be a reasonable request, that the F2PY Windows web page
 contain known combinations of version numbers for Python, Numpy and
 Gfortran verified to play nice? Some references as to queried compiler
 system environmental variables would be useful also.

I have added some numpy.distutils support for gfortran on windows, but
Windows + gfortran  + numpy is unlikely to work well unless you build
numpy by yourself with gfortran. I have actually been considering a move
to gfortran for windows builds, but I would prefer waiting mingw to
officially support for gcc 4.* series.

cheers,

David
___
Numpy-discussion mailing list
Numpy-discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion