VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Tue Feb 08 2005 - 14:00:36 CST

Hey Charles,
  Basically, the problems I'm having are quite simple. I can build
the 32-bit versions of Python on all platforms with no problems.
When I go to build Python in 64-bit mode on Solaris, IRIX, etc, I
get compile failures because the Python 'configure' scripts fail to
consistently keep the 64-bit compilations flags intact throughout the
entire build process. With Python 2.2.2 I could get around this problem
by specifying the 64-bit compile flags as part of the compiler string
like this (Solaris):
  setenv CC "cc -fast -xarch=v9a"
  setenv CXX "CC -fast -xarch=v9a"

Or this (IRIX):
  setenv CC "cc -64 -mip3 -G0"
  setenv CXX "CC -64 -mip3 -G0"

Even having to have done this for Python 2.2.2 was a hack.
The reason I had to do that was because they only used the
CFLAGS and CXXFLAGS in certain parts of the build, ignoring them
in others. The problems come in during linking of shared objects,
the main python interpreter, and various sub-parts of the overall
build, where they fail to keep the necessary 64-bit flags during
the linking phase of building module .so files and such.

The behavior of their configure script is broken to an even greater
degree if you have the misfortune of wanting to use these compilation
flags in particular:
  --without-gcc --without-cxx

If you enable either of those flags, they completely ignore CFLAGS
rewriting it with their own stuff, so you'd essentially have to go
and hand-edit the files produced by configure to fix it up.

If necessary I might be willing to do this, but it's ridiculous, and
defeats the purpose of having an autoconf script in the first place...
I'm planning to report these bugs to the Python sourceforge page,
but I still haven't got my bug reporting account active, despite having
applied a few days ago.

  John

On Tue, Feb 08, 2005 at 12:11:21PM -0500, Charles_at_Schwieters.org wrote:
[...]
> what kind of errors are you seeing? Aside from Irix (which is sorta long
> in the tooth these days), I haven't noticed problems building 2.3.4
> under Solaris, OSF, AMD64, Itanium. It's possible that specific modules
> failed to build, but I haven't noticed problems...
>
> Charles
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.1 (GNU/Linux)
> Comment: Processed by Mailcrypt 3.5.8 <http://mailcrypt.sourceforge.net/>
>
> iD8DBQFCCPK5PK2zrJwS/lYRAm0/AJ9wZtSY9aNZyevIwa3JcESL94My3ACeKeFs
> wkfG2Hn5rtpKyzGOpUmC6TY=
> =9cdD
> -----END PGP SIGNATURE----- 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349              
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078