[Axiom-legal] RE: Aldor written in C?

[Page, Bill]

On Thursday, October 13, 2005 10:23 AM Martin Rubey wrote:
> 
> Camm Maguire writes:
> If aldor can ever meet this standard, I would be happy to try to
> do the same for it, though I understand securing such permission
> might be difficult.  BTW, what language is aldor written in?
> 
> As far as I know, it's plain C.
>

Aldor is written in a combination of C and Aldor.

According to:

http://www.aldor.org/docs/reports/ukqcd-2000/compiler1-ukqcd00.pdf#page=
5

the compiler consists of about 350,000 lines of code (not sure if any
is Aldor) and from:

http://www.aldor.org/docs/reports/ukqcd-2000/compiler2-ukqcd00.pdf#page=
10

that the runtime system consists of:

25,000 lines of C code plus
 2,400 lines of Aldor code
 
> By the way, is there any progress on the licensing side?
> 

I am disappointed to report that I have had no reply from Steven Watt
in spite of several emails and a telephone call over the last two
months. :( I asked him specifically for a copy of the source license
agreement. I am beginning to wonder whether such a document actually
exists...

The terms of the license agreement to which we do have access at
http://www.aldor.org is quite permissive and the intention seems
largely compatible with GNU GPL since it requires full disclosure
of any modifications of Aldor to the original authors, although no
where is there a guarantee of access to the *full* source code -
only the Aldor libraries.

I am still convinced that Aldor is very important to the future of
Axiom (and vice versa). I also believe that it would be completely
unrealistic to think that we have the resources to developer a compiler
similar to Aldor starting from the published specifications alone,
or even to make sufficient enhancements to SPAD to support this
additional functionality.

So what next? Tim Daly has stated his opinion (I hope I am not
exaggerating his words? :) that it is likely that the licensing
agreement between aldor.org (Steven Watt) and the original developers
(IBM and NAG) is not exclusive. In principle this would allow the
Axiom Foundation to approach the original developers again for the
rights to include the Aldor source code in the Axiom open source
distribution. I do not have any indication from Steven Watt about
this possibility, however if this is not with the full co-operation
of aldor.org, then it would in effect establish a "fork" of the
Aldor project specifically for Axiom.

Forming a new Aldor licensing agreement with the original
developers is likely to take some time unless we can find
someone closely connected to these organizations who is willing
to lead such an effort. (I am thus disqualified.)

It is even possible that some of the Aldor developers who do
already have access to the source code, might be in a position
to simply establish such a fork of Aldor for Axiom by fiat on
less there exists a clear license agreement or non-disclosure
agreement to the contrary. If the problem of obtaining a source
license agreement for Aldor is simply due to the lack of time
and/or desire of Steven Watt to administer the Aldor licensing
formalities, then perhaps he might even look favorably on a
proposal to fork Aldor or to incorporate Aldor and it's special
licensing conditions into the current Axiom distribution.

I think the time has come to consider this last alternative
more seriously.

Regards,
Bill Page.