Discussion:
[sage-devel] Re: implementation of Rencontres numbers
Ralf Hemmecke
2007-12-04 15:04:55 UTC
Permalink
Hi Dan,
I've been spending a bit of time with symmetric functions recently.
In the near future, I'll be making a commit that adds support for
Hall-Littlewood, Macdonald polynomials, and quasisymmetric functions.
Other stuff that will need to be done are LLT polynomials, k-Schur
functions, and noncommutative symmetric functions.
That sounds interesting. We are also planning support for symmetric
functions in Aldor-Combinat, but haven't had the time to actually work
on it.
Also, there isn't any support for tree-like structures. I think that
implementing combinatorial species, while taking a fair amount of
initial work, would be the best way to go about this. I've started
porting Aldor-Combinat over to Sage, but that's still in the early
stages.
Mike, could you point me to your code? I actually wonder why you would
reprogram it in Sage and not interfacing the aldor-combinat library?
Is there a problem with the license of the aldor compiler
(aldor-combinat itself is GPL2).

(Sorry that I haven't been reading much on the Sage-devel list.)

Ralf Hemmecke

-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From the desktop to the data center, Linux is going
mainstream. Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
Michael Abshoff
2007-12-04 16:13:37 UTC
Permalink
Post by Ralf Hemmecke
Hi Dan,
I've been spending a bit of time with symmetric functions recently.
In the near future, I'll be making a commit that adds support for
Hall-Littlewood, Macdonald polynomials, and quasisymmetric functions.
Other stuff that will need to be done are LLT polynomials, k-Schur
functions, and noncommutative symmetric functions.
That sounds interesting. We are also planning support for symmetric
functions in Aldor-Combinat, but haven't had the time to actually work
on it.
Also, there isn't any support for tree-like structures. I think that
implementing combinatorial species, while taking a fair amount of
initial work, would be the best way to go about this. I've started
porting Aldor-Combinat over to Sage, but that's still in the early
stages.
Mike, could you point me to your code? I actually wonder why you would
reprogram it in Sage and not interfacing the aldor-combinat library?
I am under the impression that he started coding before Aldor become
"semi-free" - see below, but I guess he needs to tell his side of the
story.
Post by Ralf Hemmecke
Is there a problem with the license of the aldor compiler
(aldor-combinat itself is GPL2).
Yep, I believe so. From the website; Version 1.1 of Aldor has been
released in source form under the Aldor Public License 2.0. This allows
free non-commercial use, modification and re-distribution of Aldor
software

This is only my personal opinion: "This allows free non-commercial use"
seems unacceptable to the principles of Sage. Am I mistaken or does this
not apply to the aldor compiler itself?
Post by Ralf Hemmecke
(Sorry that I haven't been reading much on the Sage-devel list.)
;)
Post by Ralf Hemmecke
Ralf Hemmecke
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-***@googlegroups.com
To unsubscribe from this group, send email to sage-devel-***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
Ralf Hemmecke
2007-12-05 15:59:06 UTC
Permalink
Post by Michael Abshoff
Post by Ralf Hemmecke
Is there a problem with the license of the aldor compiler
(aldor-combinat itself is GPL2).
Yep, I believe so. From the website; Version 1.1 of Aldor has been
released in source form under the Aldor Public License 2.0. This allows
free non-commercial use, modification and re-distribution of Aldor
software
This is only my personal opinion: "This allows free non-commercial use"
seems unacceptable to the principles of Sage. Am I mistaken or does this
not apply to the aldor compiler itself?
Well, "free non-commercial use" is incompatible with the GPL. The APL2
is incompatible with the GPL. Sage is GPL.

I know that this causes endless problems of getting more people
interested in Aldor. Unfortunately, I cannot make Aldor truely free.
There are other people who must do it.

However, what we do in Aldor-Combinat cannot be done so elegantly in any
other language I know off. We learned a lot from MuPAD-combinat, but our
design is completely different from MC, since Aldor is so powerful.

Ralf

-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From the desktop to the data center, Linux is going
mainstream. Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4

Mike Hansen
2007-12-04 16:57:03 UTC
Permalink
Hello,
Post by Ralf Hemmecke
Mike, could you point me to your code?
I've attached the _very_ rough version I started awhile back. It just
does FormalPowerSeries and DataStream.
Post by Ralf Hemmecke
I actually wonder why you would
reprogram it in Sage and not interfacing the aldor-combinat library?
Is there a problem with the license of the aldor compiler
(aldor-combinat itself is GPL2).
Well, tree-like structures ( and hence species ) should be a standard
part of Sage. If we were to use aldor-combinat to do those
computations, then aldor would need to become a standard part of Sage.
This presents a number of issues:

* Aldor must build cleanly on all the platforms Sage builds on. This
is something that someone much more familiar with Aldor. The first
step would be to create an experimental spkg.
* I have no idea how (if possible) to interface an Aldor program from
C. pexpect interfaces are slow.
* licensing issues need to be sorted out since the APL2 is not
GPL-compatible. I don't know if it'd even be possible to distribute
Aldor as part of Sage.

Also, part of it was just a chance for me to learn more about species.
I definitely cannot program in Aldor, and one of the reasons why
Python has been so nice to work with is that is just sort of stays out
of the way. I'd much think about programming math instead of
programming languages.

--Mike

--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-***@googlegroups.com
To unsubscribe from this group, send email to sage-devel-***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
Mike Hansen
2007-12-04 17:36:14 UTC
Permalink
Post by Mike Hansen
Post by Ralf Hemmecke
Mike, could you point me to your code?
I've attached the _very_ rough version I started awhile back. It just
does FormalPowerSeries and DataStream.
Just for clarification, the code that I posted above is just some very
preliminary experimentation I had been doing on my own machine. It
isn't documented at all. I had wanted to see how smoothly ideas in
Aldor translated to Python. It is based completely off of the work
done by Ralf Hemmecke and Martin Ruby found here:
http://www.risc.uni-linz.ac.at/people/hemmecke/aldor/combinat/

--Mike

-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From the desktop to the data center, Linux is going
mainstream. Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
Michael Abshoff
2007-12-04 18:20:34 UTC
Permalink
Post by Mike Hansen
Hello,
Hello,
Post by Mike Hansen
Post by Ralf Hemmecke
Mike, could you point me to your code?
I've attached the _very_ rough version I started awhile back. It just
does FormalPowerSeries and DataStream.
Post by Ralf Hemmecke
I actually wonder why you would
reprogram it in Sage and not interfacing the aldor-combinat library?
Is there a problem with the license of the aldor compiler
(aldor-combinat itself is GPL2).
Well, tree-like structures ( and hence species ) should be a standard
part of Sage. If we were to use aldor-combinat to do those
computations, then aldor would need to become a standard part of Sage.
* Aldor must build cleanly on all the platforms Sage builds on. This
is something that someone much more familiar with Aldor. The first
step would be to create an experimental spkg.
Aldor seems to build on a reasonable number of platforms, but the binaries
are for quite old systems and from past experience exposing even
experimental packages to all the different build tool chains causes no end
of problems. The gcc 4.2 was pretty rough, gcc 4.3 promises even more
problems and we are currently working with snapshots to get on top of
that. Add platforms like gcc 4.2/Itanium and things tend to become even
more complicated.
Post by Mike Hansen
* I have no idea how (if possible) to interface an Aldor program from
C. pexpect interfaces are slow.
It doesn't seem too hard, see http://www.aldor.org/docs/HTML/chap19.html
for examples.
Post by Mike Hansen
* licensing issues need to be sorted out since the APL2 is not
GPL-compatible. I don't know if it'd even be possible to distribute
Aldor as part of Sage.
Yep, that is the key point, and I doubt that the licensing issue will be
resolved any time soon. As far as I can tell from the discussions I have
read on OpenAxiom-devel, the axiom mailing list and occasional posts by
Bill Page related to the whole issue on sage-devel it seems unlikely that
the license will chance any time soon. It was my impression that everybody
wanted Aldor to be truly free, but that the copyright owners went another
way. Feel free to correct me on that.
Post by Mike Hansen
Also, part of it was just a chance for me to learn more about species.
I definitely cannot program in Aldor, and one of the reasons why
Python has been so nice to work with is that is just sort of stays out
of the way. I'd much think about programming math instead of
programming languages.
--Mike
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-***@googlegroups.com
To unsubscribe from this group, send email to sage-devel-***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
Martin Rubey
2007-12-05 08:28:17 UTC
Permalink
Post by Michael Abshoff
Post by Mike Hansen
* licensing issues need to be sorted out since the APL2 is not
GPL-compatible. I don't know if it'd even be possible to distribute Aldor
as part of Sage.
Yep, that is the key point, and I doubt that the licensing issue will be
resolved any time soon. As far as I can tell from the discussions I have read
on OpenAxiom-devel, the axiom mailing list and occasional posts by Bill Page
related to the whole issue on sage-devel it seems unlikely that the license
will chance any time soon. It was my impression that everybody wanted Aldor
to be truly free, but that the copyright owners went another way.
I believe it would make *a lot of* sense to forward such "questions" to the
people that are most likely to be able to correct them. At least some of the
copyright owners seem to care about aldor being used, but I have the impression
that they believe that the current license works alright.

In case it helps: to *use* aldor-combinat, you actually do not necessarily need
the aldor compiler. It is sufficient to have (some version of) axiom. Of
course, you do need the aldor compiler to build aldor-combinat.

Martin



-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell. From the desktop to the data center, Linux is going
mainstream. Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
Michael Abshoff
2007-12-05 12:42:27 UTC
Permalink
[somehow Martin's reply didn't make it onto sage-devel (yet)]
Post by Martin Rubey
Post by Michael Abshoff
Post by Mike Hansen
* licensing issues need to be sorted out since the APL2 is not
GPL-compatible. I don't know if it'd even be possible to distribute
Aldor
Post by Mike Hansen
as part of Sage.
Yep, that is the key point, and I doubt that the licensing issue will be
resolved any time soon. As far as I can tell from the discussions I have read
on OpenAxiom-devel, the axiom mailing list and occasional posts by Bill Page
related to the whole issue on sage-devel it seems unlikely that the license
will chance any time soon. It was my impression that everybody wanted Aldor
to be truly free, but that the copyright owners went another way.
Hello Martin, Ralf,
Post by Martin Rubey
I believe it would make *a lot of* sense to forward such "questions" to the
people that are most likely to be able to correct them. At least some of the
copyright owners seem to care about aldor being used, but I have the impression
that they believe that the current license works alright.
[sorry for the long rants in advance]

Well, I assume the copyright holder is still NAG and they do have a bunch
of lawyers that should be aware that the license they choose is GPL
incompatible. Open sourcing projects half way or in name only doesn't
serve anybody. Aldor is a specialized piece of software that seems to have
its own niche, but it is far, far away from being main stream. So the
combination of the half-assed license together with the vast number of
truly free alternatives make it appear to *me personally* that Aldor will
languish for a while longer, but it certainly isn't revitalized by the
license change.

To get back to your original point: I don't use Aldor and/or Axiom and
because there are better, truly free alternatives out there it doesn't
make a difference for *me personally* what happens to Aldor. It is a shame
that NAG couldn't do the right thing, but know knows if there isn't some
legal reason that prevents them from open sourcing it properly. Feel free
to forward this to the "right" people, but I seriously doubt that it will
have an impact. Aldor should have been freed *many years* ago to make an
impact, but sprinkling the [pseudo] open source pixie dust on some dying
piece of code doesn't magically ressurect it. Think of Mozilla: it took 3
years to get it in shape with a mass of volunteers and a lot of financial
backing to become somewhat usable. And it barely made it in my opinion, if
a real open alternative would have shown up in that timeframe it might
have been doomed. Version 3 of firefox will be the first truly usable
version of the gecko engine. The V2->V3 transition [I believe gecko
1.8->1.9] fixed *300* memory leaks - the main issue people have been
complaining about for ages. I guess I made my point here.
Post by Martin Rubey
In case it helps: to *use* aldor-combinat, you actually do not necessarily need
the aldor compiler. It is sufficient to have (some version of) axiom. Of
course, you do need the aldor compiler to build aldor-combinat.
Ok, but no standard component of Sage will ever depend on a non-free tool
chain to build, especially one that prohibits commercial activity. I read
the Aldor license and one potential source of funding for Sage is support
and that is clearly prohibited by Aldor's license.

Another issue is that we do not want low level functionality to depend on
pexpect interfaces, i.e. Robert Bradshaw rewrote some low level operation
in Cython instead of using GAP and the speed-up was a factor of 4,500 [not
a typo]. Since one can use Aldor via C this can be worked around on a
technical level, but the licensing issue prevents us from doing so. We
also want library interfaces and while that might be possible with FriCAS
build on top of ecls that hasn't happened yet and as long as nobody
actively writes and pushes the code nothing will happen. It would also
mean that maxima needs to be ported to ecls since we will not have two
lisp implementations in tree and that hasn't happened yet either, mainly
because we lack technical expertise in that area and there are bigger fish
to fry elsewhere.

There is also the problem that we need to balance the increase in size
with the increase in functionality and build time and including FriCAS and
Aldor to get aldor-combinat seems like a bad trade off at the moment.
Obviously that doens't mean that it can't become an optional spkg [like
the openaxiom.spkg by Bill Page, maybe it should be integrated there], but
the chances of some form of Axiom to become standard are slim. We are also
pushing for Sage to become part of Debian and the Debian people really
dislike non-free software.
Post by Martin Rubey
Martin
Cheers,

Michael



--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-***@googlegroups.com
To unsubscribe from this group, send email to sage-devel-***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
Loading...