Re: [ANNOUNCE] XML/Ada 0.5 released

[Emmanuel Briot <briot@gnat.com>]

(I am posted this message on behalf of Robert Dewar for the second time.
 For some reason, the first time never made it to the newsgroup. Sorry
 if it might seem a little bit out of context, especially since I
 am not directly replying to the intended message, and thus am breaking
 the threading -- Emmanuel)


A note on the licensing of the XML libraries

The XML libraries are currently licensed under the GPL, allowing free
use in free software, but deliberately limiting use in software with
restricted distribution. This is intentional, and there are two reasons
for this. First these are beta libraries and it is entirely premature
for anyone to assume that the interfaces are fixed and to incorporate
them into any mission critical software.

Second, we are for this library inclined to license and distribute
this library in such a manner that they can be freely used by our
commercial paying customers, and also freely used by writers
of free software and students and others learning Ada.

This emphasizes that these are the constituencies to whom GNAT is
addressed. It is true that it may discourage some use in non-free
software, but we hope it may also encourage some people to consider
writing free software instead of non-free software. True, a byproduct
is that it makes it impossible for non-customers to use the library
to create proprietary products, but we have never targetted this
constituency in any case.

Ted said

> I was around before that exception existed. What happened was that someone
> piped up saying that their lawyers wouldn't allow them to use Gnat for their
> projec as they were of the opinion that linking GPL-ed packages into their
> program would require them to make the whole program GPL-ed.

This is what Winston Churchill would call a "terminological inexactitude"
[my first dratf said "complete and utter nonsense" :-)]. The original NYU
contract with the government required that the compiler runtime library be
licensed using the LGPL (because that is what I suggested to Chris), and
the reason was that we wanted the GNAT compiler itself to be as widely
usable as possible. Of course everyone involved understood what the GPL
meant. We did not need to wait for some miscellaneous lawyers to give
the (obviously correct) opinion that linking GPL-ed packages into a
program either requires you not to distribute the program, or to
distribute it under the GPL.

As the project developed at NYU, we discussed the matter with Stallman
and others and decided to use use the more permissive GMGPL instead of
the LGPL, because the LGPL is a pain in the neck (it assumes a rather
obsolete model of linking, does not take care of generics, and requires
distribution as object files, so it really does not achieve its objectives
very well). This change was accepted. There has never ever been a version
of GNAT with a GPL'ed runtime (before or after Ted's starting to be around)

So Ted is quite wrong here (really Ted, you should check your facts before
you post, one of the reasons I have decided to spend far less time with
CLA - I only read occasionally when someone points out a thread - is that
there is so much misinformation and ill-informed opinion expressed!).

Al Christians said

> What is the chance that any of the recipients of the GNAT-Modified
> GPLed versions of the library would allow a copy to slip into general
> circulation with the GNAT-Modified license attached?

Contrary to what was posted here, we have not decided to release any
versions of this library with a GMGPL license. The form of license to
our customers [really the only formal licenses that exist, since the
informal licenses where there is no contractual relationship are odd
legal beasts of dubious validity] has not yet been determined.

Ted said

> Why not just use the same lever that you do with the Ada compiler? You could
> claim that the only versions of XML\Ada certified to pass the XML conformance
> suites (or whatever it is they call them) are the ones you directly distribute
> to customers.

This is not a matter of "levers" at all, but rather a way of making clear
what constituencies Ada Core Technologies is addressing with both its
commercial and public releases, something that has not always been clear
to everyone in the past.

Peter Hendon said

> I have been working on an ADA implementation of an XML parser, DOM and
> tools. These would have the purpose stated above, and I nearly scrapped
> that project when I first read the announcement. Now... we'll see. It needs
> some serious (re-)work before I release it anyway. If I do, it will be GMGPL.

Well that's one good by-product of our decision already. It would have been
a shame if distributing this library had inhibited Peter from continuing his
work. It will benefit everyone in the Ada community if there are alternative
solutions to such problems. Note Peter that although the code we distribute
is copyrighted, we never protect with patents, and the important point there
is that you are not only free, but welcome, to look at our sources and borrow
any good ideas you see!

I will ask Emmanuel Briot to post this (I must say I feel that Deja did me
a favor in cutting off my connection to CLA :- I have much more time for
other activities, like improving GNAT!)

Robert Dewar
Ada Core Technologies