Re: Towards a free GNU Ada

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

Bob Leif said

<<

I have only one problem with suggestion 2.  I believe that any useful

additions to a compiler that are initially made for GNAT and are NOT funded

by ACT should be copyrighted in a form that they can be used by other Ada

vendors. This includes being incorporated into these vendors commercial

products.  The cost to the other Ada vendors should be the same as that

bourn by ACT.



In short, I believe that we small operators and independent users should

financially reimburse ACT, but at a reasonable price, and in a manner to

maintain competition in the Ada compiler business.>>



First, pele who contribute software to the world get to decide for themselves 
in what manner to do this. If someone writes a useful Ada library, they have several options:

a) public domain
b) GPL
c) modified GPL:
d) other similar licenses
e) fully proprietary code

Neither Bob Leif nor anyone else can dictate to authors how they
should treat their own copyrighted works, and if someone writes something
that you would like to use, but does not make it available to you at a fee you like,, you have no legitimate complaint, except simply not to use it.

That being siad, the abopve suggestion indicates a serious misun derstanding of what
is going on. In particularly I suspect that Bob is confusing the license fot the compiler
itself, with the license for code generated by the compiler. These are of course completely unrelated.

FOr example, i you want to write a proprietary replacement for one of the GNAT library routines, and refuse to
distribute the source for it, and charge $10,000 for it, with no right to redistribute,
then that's fine, and this unit can be used with GANT or with any other compiler by people who license it from you.. The
fact that GNAT itself is under the GPL and its runtime
is under the modified GPL has ZERO effect on your ability to use proprietary code with GANT>

Now what Bob is really worrying about is some academic using the GPL, and makingtheir product unusable for him (as a small independent vendor).

Let's see what is going on here ...

First if someone uses the GPL for such a project, they are saying that they are willing to have
other people use it, but ONLY if those other people are writing free software. Such a position is
perfectly reasonable, and one that I don't see anyone has any business criticizing, it;'s the author's choice.

Now of course if you *are* writing free software in Ada, you pretty much have to use GNAT. Why is this?
Simp[le, it is because other vendoirs are unwilling to make their runtimes sufifsufficiently free3 that they are consistent with free software requirements. All this would take is
for the vendor to say for 
theor runtime:

a) the sources are available
b) unlimited redistribution is allowed

Now, no one is forcing any vendor to do this, but if you feel that ACT has an unfair advantage in that GNAT can
be used to generate free software tware and your compiler cannot, that's a bit odd, since
it is you who have decided to place the unaccceptable (for free software) restrictions
on your software.

Note that you could make your runtime this free while still keeping
your compiler entirely proprietary --
as I note above the license on the compiler has
nothing to do with the license on the runtime.

Actually I think this whole line of reasoning is a giant red herring, people who
are writing free swoftware generally are NOT paying for support on GNAT, and would not be interested in buying expensive
proprietary tools anyway. 

If you are writing non-free software, then whether you are using GNAT or anything else, you cannot
make use of this product from the mythical GPL academic, (I say mythical,
because as far as I know, this concern of Bob's is hypothetical in any case).

Note that if you *do* want your runtime units or library units to be usable by
GNAT and by other compilers, then the modified GPL is just fine, and of course this is what ACT
uses for its runtime. As a result, it is quite possible for other vendors to make
use of these units, and at least in the past Aonix has (quite legally) distributed
some units from the GNAT runtime for use with their compiler.

If Bob is so concerned about a level playing field, then how about
suggesting that he open up his code! ACT makes all its code available, and welcomes our competitors to use the ideas in the compiler itself, and the ideas or even the code (if kept open) of the
runtime.

We think it would be best for the Ada community if all vendors code were this open, and
we practice what we preach :-)



<<sorry if this is hard to read, typing through Compuserve from
my hotel room in Tokyo leabves something to be desired, I don't see anything I have typed
for qyuiteuite a while!!!!>>

Robert Dewar
Ada Core tore Technologies