Re: fpectl

[Sébastien Boisgérault]

Good ! And thanks for the link. I had not noticed the warning "fpectl
module is dangerous" before.

I am a bit sad that the floating-point issue is disappearing from the
*active topics* list ...

Cheers,

SB

-- 
http://mail.python.org/mailman/listinfo/python-list

Re: fpectl

[Jeff Epler]

On Tue, Apr 19, 2005 at 02:05:11AM -0700, Sébastien Boisgérault wrote:
> Thanks for this answer.
> 
> Did you forward this info to python-dev ?

I created a patch on the sf tracker.  It's been responded to by several
developers.  You can read what they said there.

http://python.org/sf/1185529

Jeff


pgpzLX8Ht47YG.pgp
Description: PGP signature
-- 
http://mail.python.org/mailman/listinfo/python-list

Re: fpectl

[Sébastien Boisgérault]

Thanks for this answer.

Did you forward this info to python-dev ?

Cheers,

SB

-- 
http://mail.python.org/mailman/listinfo/python-list

Re: fpectl

[Jeff Epler]

It looks like the automatic build of the 'fpectl' module was broken
somewhere along the line, perhaps when the transition from Modules/Setup
to setup.py took place.

Once I made the change below and rebuilt, I got the fpectl module.
Furthermore, it appeared to "do something" on my Linux/x86 system:

$ ./python -c '1e200 ** 6'
OverflowError: (34, 'Numerical result out of range')
$ ./python -c 'import fpectl; print dir(fpectl); fpectl.turnon_sigfpe(); 1e200 
** 6'
['__doc__', '__file__', '__name__', 'error', 'turnoff_sigfpe', 'turnon_sigfpe']
Fatal Python error: Unprotected floating point exception
Aborted

Jeff


Index: setup.py
===
RCS file: /cvsroot/python/python/dist/src/setup.py,v
retrieving revision 1.217
diff -u -u -r1.217 setup.py
--- setup.py15 Apr 2005 20:32:39 -  1.217
+++ setup.py19 Apr 2005 00:13:15 -
@@ -400,6 +400,8 @@
 # select(2); not on ancient System V
 exts.append( Extension('select', ['selectmodule.c']) )
 
+exts.append( Extension('fpectl', ['fpectlmodule.c']) )
+
 # The md5 module implements the RSA Data Security, Inc. MD5
 # Message-Digest Algorithm, described in RFC 1321.  The
 # necessary files md5c.c and md5.h are included here.


pgpAzzv5PtAS8.pgp
Description: PGP signature
-- 
http://mail.python.org/mailman/listinfo/python-list

Re: fpectl

[tiissa]

Matt wrote:
Sébastien Boisgérault wrote:
My platform is Linux (Mandrake 10.x)
I'm assuming that means you can't use it on a Windows platform
It's harsh to say Mandrake is a Windows platform. ;o)
--
http://mail.python.org/mailman/listinfo/python-list

Re: fpectl

[Matt]

Sébastien Boisgérault wrote:
> Hi all,
>
> Can anybody tell me why I can't load the fpectl module in my Python
> interpreter:
>
> >>> import fpectl
> Traceback: ...
> ...
> ImportError: No module named fpectl
>
> My platform is Linux (Mandrake 10.x) + Python2.4, built from the
> (python.org) sources and configured with the --with-fpectl option.
>
> Any idea ? Is the Python Library Reference obsolete on this point or
> did I miss something ?
>
> I have understood from the previous posts on the subject that the
whole
> floating-point issue (or specifically IEEE 754 support) is quite
> complex
> and highly platform-dependent. Therefore I guess that I cannot take
for
> granted the expected behavior of fpectl functions. But at least I
> should
> be able to import it, shouldn't I ?
>
> Cheers,
>
> S.B.

I'd assume it's because you're on a non-Unix system.  The docs
(http://www.python.org/doc/2.3.5/lib/module-fpectl.html) say:
Availability: Unix

I'm assuming that means you can't use it on a Windows platform (or any
non-Unix environment)...

--
http://mail.python.org/mailman/listinfo/python-list

fpectl

[Sébastien Boisgérault]

Hi all,

Can anybody tell me why I can't load the fpectl module in my Python
interpreter:

>>> import fpectl
Traceback: ...
...
ImportError: No module named fpectl

My platform is Linux (Mandrake 10.x) + Python2.4, built from the
(python.org) sources and configured with the --with-fpectl option.

Any idea ? Is the Python Library Reference obsolete on this point or
did I miss something ?

I have understood from the previous posts on the subject that the whole
floating-point issue (or specifically IEEE 754 support) is quite
complex
and highly platform-dependent. Therefore I guess that I cannot take for
granted the expected behavior of fpectl functions. But at least I
should
be able to import it, shouldn't I ? 

Cheers,

S.B.

-- 
http://mail.python.org/mailman/listinfo/python-list