Hello !!!
> Hi Nathan, just for future reference, it would have helped had you replied
> below
> my questions, with some context, rather than start effectively a new message.
Got it ! :-)
>> OK, so lets add --with-gmp and --with-zlib, both of which give increased
> performance.
+1
> On a related note, do you need the static libraries? The default is to build
> the
> package with both static and dynamic libraries, but I doubt you need the
> static
> ones for Sage. Perhaps we can disable that too - it will reduce slightly the
> time to build, and reduce slightly the overall size of Sage.
>
> If so, --disable-static would seem appropriate though I have not tested this.
If it compiles and runs on one machine, I do not see why it should
fail on another one.
+1 to --disable-static
> I'll remove the ld library. You can then test it on multiple platforms. If you
> do not have an account on an OS X machine, I suggest you get one from William.
> I'm not in a position to test this code properly, but I will be able to get
> the
> package into some sort of shape.
I can test it on my office computer (Fedora), and on my laptop
(debian). I can ask a colleague to lend me his macbook (64bits).
> In which case, I will remove the comment "ATTENTION: YOU HAVE TO REBUILD THE
> SAGE LIBRARY TO GET THIS TO WORK"
+1
> I'll make a package, then you can remove what you feel is necessary from
> mip.pyx
I'll create a patch later on, or include the modifications in another
one... The MIP class is at rest these days :-)
> What you have
>
> include_dirs = ["../../../local/include/",
> "../../../devel/sage/c_lib/include/"],
>
> no doubt works, it just looks rather untidy and people do not know what you
> are
> referring to without a lot of detective work.
>
> I've seen code like this (from a patch to matplotlib)
>
> ### FOR SAGE
> sage_inc = os.environ['SAGE_LOCAL'] + '/include/'
> sage_lib = os.environ['SAGE_LOCAL'] + '/lib/'
>
> I don't know if that sort of thing would make more sense. It would be good to
> get some input from others on this. There may be a cleaner way to write it.
It sounds clear enough to me, at least
+1 !
> So can the 'patch' directory be deleted? If this is going to be moved to a
> standard package, it would make sense to clean it up. If it can't be deleted,
> then it should at least be renamed to 'patches'.
I think it can be deleted. It will break backward compatibility with
old versions of Sage... The users will just need to update the version
they are using ;-)
> But if I delete it, then the
> bit in setup.py
>
> setup(
> cmdclass = {'build_ext': build_ext},
> ext_modules = [ Extension("sage.numerical.mipGlpk",
> ["patch/mipGlpk.pyx"],
>
> Makes no sense at all.
Indeed. This module isn't used in the current version of Sage anyway.
> Can you be specific and give me the actual commands to run to execute the
> tests.
> i.e.
>
> ./sage -t ....
sage -t -optional sage/graphs/generic_graph.py sage/graphs/graph.py
sage/numerical/mip.pyx
should be more than sufficient. If this pass, everything else will !
Note that if you do not apply patch #9230, you may get several errors
because of the TSP function. You will also get other errors at the
beginning on generic_graph, caused by graphviz if you don't have it
installed :-)
We are currently updating the #optional flags through flags #9269. If
you apply it, you can run the tests with the option -only-
optional=glpk,cbc and spare yourself tests which do not use GLPK.
Thankssssssssssss !!!
Nathann
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
