Jim Hassett wrote:
>
> >I'd suggest first looking at the products and the suppliers
> >that match your planned architecture.  There are lots of new
> >Ada compilers out there for most major architectures.  If
> >the products look good and the suppliers seem reliable, I
> >don't see the risk for Ada to be much different from
> >anything else.
>
> I agree, but there seems to be a perception among many that with the
> changed DoD policy, the Ada market is going to shrink to the point
> where there will not be adequate resources for Ada vendors to
> create and maintain good products.  We need to counter that
> impression by pointing out the many uses of Ada that have never had
> anything to do with the so-called DoD Ada "mandate", as well as
> ongoing and new use in defense and other areas. Your information
> helps a great deal.

Others may disagree, but I don't see a big connection between
"shrinking market" and "end of mandate".  The mandate was not
well enforced anyway.  The major shrinkage occurred in the
years preceding the end of the mandate, along with the declining
defense budget.

I surveyed my customers prior to the end of the mandate,
and 85% of them said they use Ada because they think it is
superior, 15% because the were required to use it.  The
implication is that dropping the mandate would have at
most a 15% hit.  In reality, I'd guess the hit was much
lower than that, and in at least some product areas, there
has not been a hit at all.  But then, I don't have a view
into the experiences of the other suppliers out there.

Anyway, that's yesterday's news.  The mandate tumbled, what,
a year ago?  And it was smelling of death for a year before
that.  Whatever damage would be done is already done, and
most of us are still here.

> >Just because a supplier provides C or Java
> >doesn't mean they are particulary stable long-term bets.
> >Look at the problems Borland had, for one example.
>
> Good point.  One concern I've run into though is that if your C++
> compiler vendor disappears, there are lots of others to choose
> from, while there are typically few Ada vendors supporting a
> particular host/target combination.

This is a fair point.  On popular processors, there
are multiple Ada vendors.  On more obscure processors,
the situation is more hit-and-miss.

But then, using an obscure processor carries its own
baggage.  Chip vendors kill low-volume architectures
all the time, and where does that leave you?  With
a C compiler on an obsolete chip.  This happened to
me when I was working on the Digital Television project
at GE in the mid-80's.  We ended up stuck with both
an obsolete chip and a really crummy C compiler, if
you'll forgive the redundancy.

[snip]

> This is exactly the kind of data I'm looking for.  Perhaps
> you could answer a few questions about some of these items,
> to help me present things accurately:
>
> By the "past three years", do you mean 1998 and the previous
> two years (as opposed to 1995-1997)?

I meant 96-98, but you could extend the trend back
further in the form of the predecessor ActivAda product
line.

>
> By your product line, I assume you mean ObjectAda for Windows,
> right?

Yes.

>
> I assume the downloads are for the "free" version of ObjectAda.
> About when did this first become available?

Around October or November of last year.  My expectations
were that there would be a flood of downloads in the first
month or two, and then it would die down to a trickle as
the market was saturated.  I was right about the former
but wrong about the latter - activity continues unabated,
and in fact has steadily increased.

>
> Do you have a reference for the Embedded Systems Programming
> data?  I did some digging, but haven't yet found the source of
> this.

Derived from a survey they conducted (not published in the
magazine - they want you to buy ad space first!)  The
reference is the "1998 Embeddded Systems Survey".

-- Dave Wood, Aonix
-- Product Manager, ObjectAda for Windows
-- http://www.aonix.com