Re: Is Ada a commercial language ? (was: SEIC News Brief...)

[kenner@lab.ultra.nyu.edu (Richard Kenner)]

In article <1996Dec3.082243.1@eisner> kilgallen@eisner.decus.org (Larry Kilgallen) writes:
>I realize the contract ACT got from DEC may have specified 7.0,
>but if that is the basis

If you'd asked this question when we were talking at the ACT booth
yesterday (or was it Thursday?), I would have told you that the 
reason was technical.  I don't know the details (but people who do
were at the booth), but it has something to do with threads packages.

However, it isn't surprising that any product being newly developed
will be based on the then-current version of the OS.  After all, it
certainly has to work with that version anyway, so developing it for
both that and an older version can be significantly extra work.

>Of course this same issue of commercial viability is why
>the Aonix Windows NT offering is only on the most popular
>of the three hardware platforms at present,

That sounds like precisely the the right decision to me!  The usage of
the other two NT platforms is *very* small.  At ACT, we haven't
received any inquiries about GNAT on PPC/NT and only a very slight
level of interest in Alpha/NT.  I imagine Aonix has heard likewise.

>and why even GNAT is not scheduled to be ported to VAX/VMS. 

"scheduled" is a peculiar word to use in this context.

Perhaps you missed it, but the issue of GNAT on VAX/VMS was discussed
at the GNUG meeting at Tri-Ada last Wednesday.  As Robert said, all
the pieces needed from a technical point of view are present.  The
question is the extent of commercial interest.

There are hundreds of different targets out there.  If you restrict
consideration to just those already supported by GCC (including
VAX/VMS), there are still nearly 300.  It's not feasible to support
more than a small fraction of that number, so choices need to be made.

Though there have been a number of inquiries about GNAT on VAX/VMS, so
far everybody (including you, by the way) has come up with a very
small number when asked about the number of seats of support they'd
need.  That may not add up to enough to be commercially viable.  On
the other hand, the cost of this port is significantly lower than many
others, since most of the pieces exist.  My own view is that this is
"right on the edge" of what's worthwhile for ACT to do, but this is
clearly something that ACT needs to further discuss internally, as
well as with Digital and its customers.

On a more general note, the question of "ophan systems" (like "orphan
drugs") is a tough one.  The 6811 has been discussed at length in this
group and other processors (such as those commonly used by segments of
the military) were discussed on the floor of Tri-Ada.  In the present
fiscal realities, it seems unlikely that any one customer would want
to foot the bill for such development and it would be foolhardy for
any company to create such a product "speculatively" in the hope of
attracting customers for it.  The result of this will be that people
who need to use such processors will have to be granted waivers since
no Ada compilers for such will exist.  This is a problem I think AJPO
will have to come up with a solution for over the next couple of
years.