On Sat, Aug 1, 2009 at 1:50 PM, Nathann Cohen<[email protected]> wrote:
>
> Hmmmm... There is still something I did not understand, sorry ;-)
>
> Coin-or/Cbc is meant to be an optional package, and if approved it
> will stay this way : fine
> GLPK should be a standard package, but as it is customary it will be
> optional for a while ( and perhaps a bit more to let us think about
> license issues ^^; )
>
> What is the future of the class numerical.MIP (cf.
> http://trac.sagemath.org/sage_trac/ticket/6502 ) which uses all of
> this but is totally useless without either Coin or Glpk ? Thinking
> Coin may not be available, this class detects whether it is installed
> and acts accordingly, but as I thought GLPK would always be installed,
> it assumes it is available without any checking.
> Will this class be included in the standard Sage, then ? And if so, do
> I need to write some code to make it detect the presence of GLPK and
> totally refuse to run otherwise ?
>
> I just finished to write the docstrings of the function edge_coloring
> ( coloring() has been renamed vertex_coloring as it is better to write
> a LP for edge_coloring than to use vertex_coloring on a line graph ).
> All the functions I mentionned are working and documented. In the end,
> when should I post a patch for them ? ;-)
>
> Thank you for your help ! :-)
>
> Nathann

I personally think GLPK will end up being standard in Sage before
long, so I think your approach right now is pretty good.   It's just
that it has to be optional for a bit (a month or so), since that's
what we do.

Make glpk optional + making your patch trivial to install, and lots of
people will then easily be able to play around with it.

 -- William

>
> On Aug 1, 7:59 pm, William Stein <[email protected]> wrote:
>> On Sat, Aug 1, 2009 at 6:24 AM, David Joyner <[email protected]> wrote:
>>
>> > I think Nathann Cohen has done a very valuable service with the GLPK and
>> > COIN-OR-related packages.
>>
>> > That said, I have a "point of order" question. Is is true or false that the
>> > process for a package to become standard we
>> > (1) use trac to do nomination, testing, and acceptance as an optional
>> > package,
>> > (2) someone (William say) posts the spkg to
>> >http://www.sagemath.org/packages/optional/,
>> > (3) after a period of a few months, voting is done for making it standard.
>>
>> > In any case, I vote +1 for Nathann's GLPK spkg being moved to optional
>> > and the currently posted GLPK spkg on experimental removed, with the
>> > idea that it be proposed for inclusion as a standard package.
>>
>> Yes, this is exactly right.   So GLPK should become optional for a while
>> before we even vote on it being standard.
>>
>> NOTE: GLPK is GPLv3.  Since we need to retain the ability to release GPLv2
>> versions of Sage for now, this is another very good reason to make it
>> optional for a while (so it is easy to swap out).
>>
>> William
>>
>>
>>
>>
>>
>> > If the above process is correct then there is nothing more to say.
>> > If not, I think  someone (not me) look into this PyGLPK (I am
>> > not an OR person and have never heard of this before). By
>> > compare, I mean, look at the license, compare Nathann's wrappers and
>> > docstrings to PyGLPK's, etc. Maybe there are other wrappers
>> > on the internet (I have not done a google search, just mentioning
>> > possibilities here.)
>>
>> > On Sat, Aug 1, 2009 at 8:55 AM, Robert<[email protected]> wrote:
>>
>> > > I'd like LPs, too, of course.
>>
>> > > By the way, do you know of the projecthttp://pymprog.sourceforge.net/
>> > > that wraps GLPK with it's modelling language in python? Using PyGLPK,
>> > > as an additional layer, I'm pretty sure there is some relevant code
>> > > there.
>>
>> > > On Aug 1, 12:46 pm, Carlo Hamalainen <[email protected]>
>> > > wrote:
>> > >> On Sat, Aug 1, 2009 at 9:41 AM, Nathann Cohen<[email protected]>
>> > wrote:
>> > >> > Are you interested by LP features in Sage with GLPK as the native
>> > >> > solver ? ( The others would have to be optional packages but we
>> > >> > thought it would be smart to have a native one ).
>>
>> > >> Yes, especially if it can be used to speed up the graph colouring code
>> > >> in some cases (at the moment it uses my C++ dlx solver but I'm sure
>> > >> that LP etc could be useful in some other cases).
>>
>> > >> --
>> > >> Carlo Hamalainenhttp://carlo-hamalainen.net
>>
>> --
>> William Stein
>> Associate Professor of Mathematics
>> University of Washingtonhttp://wstein.org
> >
>



-- 
William Stein
Associate Professor of Mathematics
University of Washington
http://wstein.org

--~--~---------~--~----~------------~-------~--~----~
To post to this group, send an email to [email protected]
To unsubscribe from this group, send an email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---

Reply via email to