new optional (or experimental) package CSDP?

[Dima Pasechnik]

Dear all, 

we'd like to add CSDP, (a fast semidefinite programming solver, one of the very best around), 

as an optional package.

https://projects.coin-or.org/Csdp/ 

It is a C library+headers, and I've written almost all what is needed (besides spkg-install etc)

to have it seamlessly installable (using autotools) into Sage, here:

https://github.com/dimpase/csdp

One major "customer" of CSDP in Sage is Flagmatic, see http://flagmatic.org/

We are also working on a Cython interface to CSDP, and hope to have it ready soon.

CSDP itself has a GPL-incompatible license imposed by the employer of the developer of CSDP, so 

it itself cannot become standard.

I understand that we need to vote on this. (Needless to say I'm committed to support CSDP as an optional

package of Sage for the coming N years).

Please vote on the following options:

1) Yes to CSDP becoming an optional package

2) no to 1), but yes to CSDP becoming an experimental package.

3) no, no, just go away.

Thanks,

Dima

[Nils Bruin]

On Apr 10, 8:52 pm, Dima Pasechnik <dimp...@gmail.com> wrote:
> 2) yes to CSDP becoming an experimental package.

Doing that requires no different work from preparing it to be an
optional package, so why not do that first? Once that's done I would
expect it'll be pretty smooth sailing into optional status.

[Julien Puydt]

Le 11/04/2013 06:32, Nils Bruin a �crit :

+1

Snark on #sagemath

[Nathann Cohen]

+1

I had never heard of Flagmatic before. Looks GREAT :-D

Nathann

On Thursday, April 11, 2013 6:56:45 AM UTC+2, Snark wrote:

Le 11/04/2013 06:32, Nils Bruin a �crit :

[Dima Pasechnik]

On 2013-04-11, Nathann Cohen <nathan...@gmail.com> wrote:
> ------=_Part_629_6688666.1365672308379
> Content-Type: text/plain; charset=ISO-8859-1
> Content-Transfer-Encoding: quoted-printable

>
> +1
>
> I had never heard of Flagmatic before. Looks GREAT :-D

it implements a range of recent ideas due to Razborov et al, which
produced a lot of progress in extremal (hyper)graph theory recently.

Dima

>
> Nathann
>
> On Thursday, April 11, 2013 6:56:45 AM UTC+2, Snark wrote:
>>

>> Le 11/04/2013 06:32, Nils Bruin a =EF=BF=BDcrit :=20
>> > On Apr 10, 8:52 pm, Dima Pasechnik<dimp...@gmail.com> wrote:=20
>> >> 2) yes to CSDP becoming an experimental package.=20
>> >=20
>> > Doing that requires no different work from preparing it to be an=20
>> > optional package, so why not do that first? Once that's done I would=20
>> > expect it'll be pretty smooth sailing into optional status.=20
>>
>> +1=20
>>
>> Snark on #sagemath=20
>>

[Nathann Cohen]

>> I had never heard of Flagmatic before. Looks GREAT :-D
> it implements a range of recent ideas due to Razborov et al, which
> produced a lot of progress in extremal (hyper)graph theory recently.

I had no idea it had evolved into an open source project already :-)

Nathann

[Dima Pasechnik]

here is the ticket adding CSDP as an experimental package:

http://trac.sagemath.org/sage_trac/ticket/14505

[Nathann Cohen]

> here is the ticket adding CSDP as an experimental package:
> http://trac.sagemath.org/sage_trac/ticket/14505

Wow. Cool !

Why do you prefer to create 2 tickets for that ? It feels a bit weird to review an interface between Sage and a spkg which is not included, and it feels weird to add to Sage a spkg which is not interfaced with it.

Though if the two tickets depend on each other, then no problem ;-)

Nathann

[Dima Pasechnik]

On Monday, 29 April 2013 21:09:44 UTC+8, Nathann Cohen wrote:

> here is the ticket adding CSDP as an experimental package:
> http://trac.sagemath.org/sage_trac/ticket/14505

Wow. Cool !

Why do you prefer to create 2 tickets for that ? It feels a bit weird to review an interface between Sage and a spkg which is not included, and it feels weird to add to Sage a spkg which is not interfaced with it.

e.g. Flagmatic only needs csdp executable installed.

So, in a way, it already makes sense without anything extra.

[Nathann Cohen]

> e.g. Flagmatic only needs csdp executable installed.
> So, in a way, it already makes sense without anything extra.

Oh. I see !

Haveyou already created a ticket for the interface ?

Nathann

[Dima Pasechnik]

not yet, it's in the works still.

 

Nathann

[Nathann Cohen]

> not yet, it's in the works still.

Ok, cool ! Send me a line when you will begin to work on that. And to
Jernej too, whom I added to this email. He may be part of GSOC this
summer, and may want to work on that too.

Could you send an email to sage-devel to ask them "how the hell we are
supposed to libtoolize things without modifying upstream" ? I don't
know either, but I still have scars from the first spkg I put in Sage
and in which I modified upstream : Cliquer :-P

Admittedly it was not a very nice piece of work, though. I mean, what
I did with it of course ! :-D

Nathann

[Dima Pasechnik]

In this case all my changes are perfectly traceable on github.
I don't see a point in packaging these in one big patch.
As we hopefully are moving towards a saner sources management model,
big patches should be dying away, IMHO...

Dima


>
> Nathann
>

[Dima Pasechnik]

In gmane.comp.mathematics.sage.devel, you wrote:
>> not yet, it's in the works still.
>
> Ok, cool ! Send me a line when you will begin to work on that. And to
> Jernej too, whom I added to this email. He may be part of GSOC this
> summer, and may want to work on that too.

The code is here:
https://github.com/mghasemi/pycsdp
(written by our new postdoc)

it still lacks docstrings, etc etc.

Dima

[Nathann Cohen]

> The code is here:
> https://github.com/mghasemi/pycsdp
> (written by our new postdoc)
>
> it still lacks docstrings, etc etc.

O_o

But how do you get this into Sage ? I don't get how you work...
Probably further on the GIT road than I can understand right now :-P

Nathann

[Dima Pasechnik]

On 2013-04-29, Nathann Cohen <nathan...@gmail.com> wrote:

>> The code is here:
>> https://github.com/mghasemi/pycsdp
>> (written by our new postdoc)
>>
>> it still lacks docstrings, etc etc.
>
> O_o
>
> But how do you get this into Sage ? I don't get how you work...

I'm telling him for a month already to get a Sage trac account, you
know...
Well, I hope it will happen, and I don't have to scream... :)
Anyhow, that's the most unpleasant part of the inteface done, creating
linked with pointers C structs to be passed to the solver.

> Probably further on the GIT road than I can understand right now :-P

no, it's nothing to do with git per se.

Dima
>
> Nathann
>

[kcrisman]

On Wednesday, April 10, 2013 11:52:30 PM UTC-4, Dima Pasechnik wrote:

Dear all, 

we'd like to add CSDP, (a fast semidefinite programming solver, one of the very best around), 

as an optional package.

https://projects.coin-or.org/Csdp/ 

I assume this works on Linux as well - there were some remarks in the flagmatic user guide that indicated not, but maybe that wasn't because of CSDP.  I guess I recall that optional packages should work on all (most?) platforms we support Sage on. 

[Thierry Dumont]

Le 29/04/2013 19:23, kcrisman a �crit :

>
>
> On Wednesday, April 10, 2013 11:52:30 PM UTC-4, Dima Pasechnik wrote:
>
> Dear all,
> we'd like to add CSDP, (a fast semidefinite programming solver, one
> of the very best around),
> as an optional package.

> https://projects.coin-or.org/Csdp/ <https://projects.coin-or.org/Csdp/>
>

A very good ide! BUT: have you seen that, on
https://projects.coin-or.org/Csdp/ the first links you can see are broken ?

>
> I assume this works on Linux as well - there were some remarks in the
> flagmatic user guide that indicated not, but maybe that wasn't because
> of CSDP. I guess I recall that optional packages should work on all
> (most?) platforms we support Sage on.
>

> --
> You received this message because you are subscribed to the Google
> Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[Dima Pasechnik]

On 2013-04-29, kcrisman <kcri...@gmail.com> wrote:
> ------=_Part_5928_32884296.1367256180205
> Content-Type: text/plain; charset=ISO-8859-1

of course my spkg works on Linux, and even on cygwin... ;)
Didn't check Solaris yet.

Flagmatic has been fixed since, too.

>

[Dima Pasechnik]

On 2013-04-29, Thierry Dumont <tdu...@math.univ-lyon1.fr> wrote:
> This is a multi-part message in MIME format.
> --------------060206000403010900080606
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> Content-Transfer-Encoding: quoted-printable
>
> Le 29/04/2013 19:23, kcrisman a =E9crit :

>>
>>
>> On Wednesday, April 10, 2013 11:52:30 PM UTC-4, Dima Pasechnik wrote:
>>
>> Dear all,
>> we'd like to add CSDP, (a fast semidefinite programming solver, one
>> of the very best around),
>> as an optional package.

>> https://projects.coin-or.org/Csdp/ <https://projects.coin-or.org/Csdp=
> />
>>
>
> A very good ide! BUT: have you seen that, on=20

> https://projects.coin-or.org/Csdp/ the first links you can see are broken ?

..."Last modified 3 years ago"

Well, the CSDP code works, and is very popular.