Re: what DOES the GPL really say?

[dewar@merv.cs.nyu.edu (Robert Dewar)]

Ronald Cole says

<<Oh, and I openly challenge your fictitious notion that free
distribution of your "wavefronts" would be harmful.  If the very
existance of freely available GPL'd (unwarrantied) software were
harmful in some way, I'm quite sure that Stallman would have written
the GPL in such a way as to allow for its nullification in those
circumstances.>>


There is a very important difference between unwarrantied sotware and
unreliable software, in fact the two concepts are pretty much unrelated,
and I think this is probably clear to most CLA readers. There is warrantied
software that is unreliable, and there is unwarrantied software that is
reliable. ACT works hard to make sure that the publicly available vesions
of GNAT fall into the second category.

Occasionally, in the context of a specific customer with support, the
use of a non-fully tested wavefront version to solve a specific problem
is justifiable, however in our view, it would most definitely be harmful
to have such versions wandering around. It would cause a lot of confusion,
and result in a negative impression of GNAT that would in our view be
harmful to the GNAT project. Some potential users of free software are
definitely concerned, because they assume that the free software scene
is as Ronald would like to paint it, with all sorts of people making
changes to the software in an uncontrolled manner.

The best way to think of ACT is to think of it like any other software
vendor who is aiming at reliable releases. We test carefully, and release
only when we think the time is right. We are no more about to distribute
our modifications without full testing than any other company. The one
difference, which as I have said many times before we are committed to
continuing, is that GNAT releases will continue to be made publicly
available. Yes, these public releases will lag a little behind the most
super duper version available at ACT in some cases -- something that
you could say about any releases of any software, but we think that
even with this small lag, you will see GNAT technology that is significantly
ahead of the proprietary competition.

Yes, we understand that the enthusiasts around (like Ronald, but it would
be a mistake to assume that all enthusisasts share all Ronald's opinions)
would like to be able to fiddle with the latest versions of everything all
the time, and do not care about reliability. However, this is not the
market place we are addressing. The market place we *are* addressing is
serious large projects needing support, and at the same time, consistent
with this primary goal, we want to support the use of GNAT in teaching
and casual use, since we think that it plays an important part in the
continuing spread of Ada 95. We do in fact put in quite a bit of effort
in making GNAT suitable for such purposes, and are delighted to see it
being used widely for teaching at many universities, and also at the
Air Force Academy.

ACT has established a set of policies for handling GNAT releases that
concentrate first and foremost on providing high reliability. The continued
availability of sources will satisfy many other needs, though we certainly
do not expect to be able to satisfy everyone's needs, and in particular,
no matter how often and loudly Ronald (or others) yell, we do not intend
to adopt a policy of freely releasing insuffiently tested software.

I think that from our point of view, enough has been said in this thread
which has not been a particularly productive one, and this will act as
our final statement in the matter. If anyone is interested in asking
specific questions about GNAT or the associated release policies, they
are welcome to follow up with email questions to me, or to sales@gna.com
as appropriate (oops that should be sales@gnat.com -- that's one word
that should be spelled correctly, even if my T key is sticking!)

Robert Dewar
Ada Core Technologies

P.S. we froze the 3.10b sources a few days ago, and they are now undergoing
final testing. The 3.10p public release will be based on 3.10b, and will
follow very shortly if no significant problems appear.