Re: glpk and GLPKMEX

[Tommaso Balercia]

On Tue, Feb 15, 2011 at 7:43 PM, John W. Eaton <address@hidden> wrote:

> In any case, I would certainly consider any patches that improved
> Octaves glpk function.

Hi John, I finally managed to find some time to work on the patch to
use all the capabilities of GLPK 4.36+ (and obviously update the
documentation).

Before I submit it here, a few stylistic points.

1) Should I update the copyright of the files I modified? How?

2) At the moment, the presence/absence of GLPK on the system is
handled at compile time, while the presence/absence of the right
version of GLPK (4.36+) is checked at runtime? Would you prefer to do
everything at compile time?

This said, at the moment, the version I created doesn't allow you to
save a separate file in as many formats as GLPKMEX lets do, and, above
all, you can specify the filename but it's hardcoded to be "outpb.lp".
This is compliant with the behavior of the existing code I updated,
but of course it's a hassle. I will fix it in due time; my priority
was to provide Octave with the possibility to handle binary problems,
because that's pretty much the only type of linear problems I work
with on a daily basis.

Let me know about 1) and 2).

Kind regards,

Tommaso