RE: [Axiom-legal] Licensing Aldor (was: GPL vs. modified BSD)

[Bill Page]

On November 21, 2006 10:10 AM Frederic Lehobey wrote:
> 
> On Tue, Nov 21, 2006 at 09:19:43AM -0500, Bill Page wrote:
> > 
> > Aldor is "almost free" (perhaps you will claim that there is no
> > such thing?). Many people have copies of the Aldor source and
> > binary versions are publicly available.
> 
> I do not have any. And am I not interested in having it *as it is
> not free software* (what could I do from it, not having enough
> interesting rights).

You could compile your own binary for the hardware of your choice.
(I did this for x86-64 bit linux.) You could read and learn more
about the compiler. You could fix bugs in the compiler for your own
use. Only (apparently) you cannot re-distribute it to others. Although
that seems to depend on how you interpret the Aldor Public License.

> 
> Note that in the past (1999) when Axiom was "almost free" (we had
> access to all the libraries but not the source of some binary parts
> in the core system) I have been severely beaten (during the last
> months of my PhD) by some (faulty) binary only parts. At that time
> I have sweared (thanks to my Axiom very bad experience) to ever use
> again only *really* free software. Please understand why I am so
> picky about freeness of anything. It is not a goal. But a prerequisite
> from where anything is possible.

Yes, I understand your point of view and the reasons.

> 
> See how much time you are wasting with non-free Aldor.
>

I believe I would have to waste much more time with free SPAD.
 
> > > I have been constantly contradicted by people like you (and
> > > Ralf if I do not mistake) who where 'waiting for Aldor', the
> > > 'future' of Axiom.
> > 
> > Actually Aldor is the past of Axiom. Aldor was developed by the
> > same IBM research group that developed Axiom. And Aldor (then
> > named AXIOM/XL) was a part of Axiom the last time Axiom was
> > available as a commercial product. And a significant amount of
> > work has already been done to convert the Axiom library to Aldor.
> 
> Have you used Aldor at that time?

Yes in 1999 I used the Aldor compiler that was part of the Windows
version of Axiom that was distributed by NAG. However I only had a
copy of the promotional Cdrom which contained a time-limited license
manager. I was comparing it to Maple which I already knew fairly well.
I nearly purchased Axiom at that time from NAG but compared to Maple
Axiom seemed too slow. (On common hardware available today and the
changes in Maple since that time I am sure that it would now seem
much faster than Maple.) But I really liked the Axiom language and
the Windows TeX browser interface.


> I have been waiting from 1995 to 1999 for Aldor to become usable
> for research which it has never achieved (and by far).

I think you just missed the time.

> Aldor might be a funny pet project (as could be Mathemagix or SAGE)
> that could grow into something interesting (as is SAGE today).

Aldor was packaged commercially with Axiom by NAG. I do not think
it is accurate to describe it as a "pet project".

> But as it is not free software, I DO NOT CARE (and I deeply regret
> it is favoured instead of Axiom old compiler in the Axiom community).
> It is now too late for Aldor. It has missed its time frame and
> opportunity (now filled by SAGE).

Maybe. As you know I am also very interested in SAGE. I wrote
an interface for Axiom in SAGE. SAGE uses Python as the main
development language and that has some big advantages as well as
a few drawbacks. I am concerned however that the SAGE developers
are not learning as much from past experience as they should. As
a result they are re-inventing some wheels (despite their motto)
and repeating some of the same mistakes. If they had open access
to Aldor I think it might be possible to influence some of their
design choices.

Still, I think SAGE is an exciting project and the AXIOM project
should plan to work with them. I agree that SAGE sets a good
example by insisting that every part of SAGE is open source.

> I would be sad it were the same for Axiom because everybody has
> its eyes (wrongly) focused on Aldor. Maybe it is already too
> late...

No, there are some Axiom people who are not waiting for Aldor.

> 
> Many "failed" projects during their commercial life turn into
> free software (this is what occurred to Axiom). Maybe Aldor is
> not failed enough to become free software (but, ironically, I
> believe you actually never had so much chance to have Aldor
> finally released as free software).
> 

You mean you do not believe the current owners of Aldor that
they want to make in free?

> > But you are wrong that I am against the improvement of SPAD.
> > I just believe that it is too difficult a problem for a
> > volunteer-based open source project. Apparently it was even
> > too difficult for the original Axiom developers. That is why
> > they wanted to develop Aldor.
> 
> I am actually sad we have not driven enough lisp gurus to fix
> the compiler. But (rewriting a compiler from scratch is an
> industrial project the (free) Axiom community is far from being
> able to consider.

I agree. When you compare Lisp to Python (as used in SAGE) - not
the languages but rather the community), then it is clear that
Lisp is part of the problem. When I spoke to SAGE developers about
Axiom and I mentioned Lisp their eyes "glazed over" and then smiled
until I pointed out that when Axiom was originally developed Lisp
was about as old as Python is now during the development of SAGE.

> ... 
> > > For your own information, I have looked into the old compiler
> > > to see if I was able to fix things I consider critical (being
> > > able to have variables instead of constants in domains and
> > > category calls, something Aldor is able to do).
> > 
> > Is Aldor really able to do that? Wouldn't that violate the
> > principle that types are static?
> 
> It is supposed to. What I mean is being able to create domains in
> compiled code with variables in their parameters (not change the
> value of the variable after instantiation). With this much of the
> Galois field factorisation of Axiom would be much easier to write
> *and maintain* (currently, it is a kind of hack using the integers
> and contradicting the grounding principles of Axiom).

I think you could expect a lot of help on this from Ralf Hemmecke
and Martin Rubey. I am not entirely sure what you mean by "variables
that do not change value after instantiation". I think this is the
definition of a constant. So what you want to do should be possible.

> ... 
> I think it boils down to what Tim has summarized better than I
> could. It is only a matter of respect. Respect THEIR choice and
> draw you own conclusions from it.

My conclusion is that their choices (or actually rather lack of
choice) has had a very negative impact on Axiom. I want them to
stop this and license Aldor properly.

> 
> In French we have < un tiens vaut mieux que deux tu l'auras >
> which sounds like "better now than tomorrow" and "promises are
> for those willing to believe them". The reality is what we really
> have. The rest is literature.

Yes, I am sympathetic to your point of view but for me right now
this is too conservative. I want more and it is close enough that
I feel it.

> 
> > To me it is a matter of making best use of (potentially)
> > available intellectual investment. Right now both Axiom and
> > Aldor are suffering because what was meant to be together was
> > "sold" separately.
> 
> This is where we differ. You are taking into account potentialities
> (which make you not to move on the old compiler front) whereas I am
> only considering reality.
> 
> By waiting long enough, you increase your chances to finally be
> right. But the delay might be much too long to go anywhere on the
> road to the 27 year horizon. (Time flies, you know?)
> 

Indeed! :-)

Regards,
Bill Page.