Re: relax on Mac with fink -- February 26, 2010 - 16:46

Thanks for that information.  I have seen similar python specific
problems (v2.6 vs. v2.5) with 64-bit x86_64 architectures with Linux
and Windows, so I am interested in incorporating all of these changes
and using this for myself to release python specific relax versions
for Linux and Windows.  I've now made a change to the sconstruct
script which removes the need for this patch.  See
https://mail.gna.org/public/relax-commits/2010-02/msg00344.html and
http://svn.gna.org/viewcvs/relax/1.3/sconstruct?rev=10948&view=diff&r1=10948&r2=10947&p1=1.3/sconstruct&p2=/1.3/sconstruct.
 The version specific python binary now is given by py_bin_fullpath
and can be used on other systems.

Cheers,

Edward


On 26 February 2010 15:18, Jack Howarth <howarth@xxxxxxxxxxxxxxxx> wrote:
> Edward,
>   I forgot that to mention that I also needed to use this patch as well..
>
> --- relax-1.3.5/sconstruct      2010-02-25 08:58:20.000000000 -0500
> +++ relax-1.3.5.patched/sconstruct      2010-02-26 09:09:52.000000000 -0500
> @@ -500,7 +500,7 @@
>
>         # Catch Mac OS X and send the correct command line options to the 
> linker (these may become redundant as SCons improves).
>         if env['PLATFORM'] == 'darwin':
> -            env.Append(LINKFLAGS = ['-bundle', '-bundle_loader', 
> sys.prefix + path.sep + 'bin' + path.sep + 'python', '-dynamic', 
> '-undefined', 'dynamic_lookup'])
> +            env.Append(LINKFLAGS = ['-bundle', '-bundle_loader', 
> sys.prefix + path.sep + 'bin' + path.sep + 'python2.X', '-dynamic', 
> '-undefined', 'dynamic_lookup'])
>             env['SHLINKFLAGS'] = SCons.Util.CLVar('$LINKFLAGS')
>
>         # Shared library prefix and suffix.
>
> ...to make sure the correct fink python is linked against. FYI, in fink we 
> currently support
> python 2.5 and 2.6. The convention is for programs to directly access these 
> by their
> versioned names (although a python package is provided for whatever is 
> considered the
> current 'default' for fink to handle those programs to broken too use a 
> versioned python).
>   As for building with scons using a particular python, I execute scons 
> with the
> command...
>
> %p/bin/python%type_raw[python] %p/bin/scons
>
> in relax-py.info. I really wouldn't worry about these remaining fink
> patches as they are trivial to apply in relax-py.patch during the fink
> build of relax-py.
>           Jack
>
> On Fri, Feb 26, 2010 at 09:54:19AM +0100, Edward d'Auvergne wrote:
> > This is awesome news!  Now the test suite is passing 100% on all major
> > platforms (it should also pass for solaris, the BSDs, etc., anywhere
> > Python and the python modules are installed).  As for python specific
> > change,
> >
> > diff -uNr relax-1.3.5/relax relax-1.3.5.patched/relax
> > --- relax-1.3.5/relax 2010-02-25 08:58:20.000000000 -0500
> > +++ relax-1.3.5.patched/relax 2010-02-25 09:01:26.000000000 -0500
> > @@ -1,4 +1,4 @@
> > -#! /usr/bin/env python
> > +#! /usr/bin/env python2.X
> >
> >  ###############################################################################
> >  #                                                                         
> >     #
> >
> > I'm guessing you have python2.6 in the final version?  I like these
> > ideas for the released code as the C modules are often specific to the
> > exact Python version there were compiled against.  I've been thinking
> > for a while about providing v2.5 and v2.6 compiled relax versions (for
> > GNU/Linux and MS Windows).  But how did you get scons to target a
> > specific Python version?
> >
> > Cheers,
> >
> > Edward