Re: Win32Ada

[dewarr@my-dejanews.com]

In article <364EA0F6.A66B888F@cts.com>,
  dpw@cts.com wrote:
> dewar@gnat.com wrote:
>
> [Needless to say, nothing derogatory was intended
> here - I was talking about the original GNAT
> investment, not whatever proprietary follow-on
> has gone on since by ACT or anyone else.  By contrast,
> to my knowledge Win32Ada was 100% taxpayer funded, and
> IMHO ought to be a public resource.  Allowing rights
> to an uninvolved 3rd party (Microsoft), seems completely
> nutty to me.]
>

The government often funds under agrements that will leave
entire proprietary rights in the hands of companies. Indeed
my understanding is that the academic edition of Object
Ada, in which the taxpayers invested directly more than
half what it spent on GNAT, is in this category. That is
by no means unusual, and by no means necessarily
inappropriate. It really depends on what the government
hoped to achieve.

In the case of GNAT, it was of the essence in terms of the
governments interest (to make a freely available high
quality Ada 95 compiler available for academic use) to
insist on the GPL licensing of the initial version (and
that the copyrights be assigned to the FSF).

In the case of the compiler work funded later as the
academic Ada compiler project, the government made a
specific decision that there was no requirement that it
should be freely distributed in OSS form.

I think what happened in the Intermetrics bindings cases
was that the government DID intend that the resulting
bindings be freely available, but clearly did not properly
write this into the contract as they did with GNAT. The
GNAT contract is quite remarkable in that it includes the
entire text of the GPL, and very specifically required the
use of the GPL, and the assignment of the copyright to the
Free Software Foundation.

This was in my view a significant error in the contractual
instruments for this work. The other major oversight was
the failure to provide for continued maintenance.

We are now investigating the possibility of a clean room
implementation of a thin binding to Win32 that will once
and for all clear up this unfortunate confusion. This is
incidentally exactly the sort of project that one would
hope can be successfully achieved using the OSS model. The
copyright on the Win32 bindings is particularly unfortunate
from this point of view, since it is one thing for
companies like Aonix and ACT to work out how they can
distribute the Win32 bindings in their current copyrighted
form, and quite another for volunteers to distribute
modified and corrected and improved versions.

By the way, I certainly did not take anything Dave said as
derogatory, I just wanted to make the GNAT funding
situation clear. We still run into people who think that
the government is directly supporting GNAT, a situation
that of course has not been true for four years. We did not
even seek such continued funding after the initial contract
ended, since we felt that the commercial support model was
more appropriate.

There is no doubt that there could not have been a GNAT
without the government's initial funding (indeed even with
that funding, the fate of GNAT was in the balance a few
times when the project was severely attacked by some
existing Ada vendors). I think it is quite a reasonable
model for the government to provide initial funding of this
type, especially if the result is a freely available
product. I do NOT think it is healthy for the government to
continue to provide such funding.

The future of GNAT depends on our ability to improve the
product and compete directly with the other Ada 95 vendors.
This competition is an important factor in the continued
development and improvement of GNAT (not to mention the
continued development and improvement of other vendors
products!)

> I don't want to get into a protracted discussion
> on this since I have no special expertise in it
> and you are doubtless one of the top-ranking
> experts, but my gut instinct is that if something
> is 100% funded by tax dollars, it ought to be 100%
> in the control of the taxpayers or of their elected
> representatives, at least if the creator of the
> product has no intention to support and evolve it.
> I can see that an exception might be a flat-out
> grant where a priori there are no strings attached,
> such as for basic research.

I don't think that's right. If the government funds
something at a 100% level, they have a choice as to who
ends up with the data rights. Clearly they will have to
spend more if they want the data rights to be freely
available to the public. Whether this is a good idea (to
spend this additional money) depends on the project. The
ATIP projects funded a few years ago, quite deliberately
allowed the data rights to remain with the vendors, and
most of these products remain proprietary (but not all,
for example the GNAT version for the MAC was distributed
freely, and is still freely available)

> I'm reminded of ALS, a big fat waste of tax money if
> ever there was one (and how many of us had at least
> some peripheral involvement with THAT mess?)  If I
> remember correctly, the government retained control
> of the resulting software and made it available on
> mag tape for something like $25, more or less cost
> of materials and shipping/handling.

ALS cost the government a LOT of money, something of the
order of 15-20 times the cost of GNAT. Interestingly there
was much less hue and cry from the other vendors, I think
simply because ALS was seen as a technical failure, and was
never successful in the market place. Part of the reason
for some of the vendors strongly opposing the GNAT project
was their prediction (accurate we believe :-) that GNAT
would be of sufficient quality to seriously compete with
their products.

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own