Re: Ada Compiler Pricing (was Re: Hijacking a Thread was RE: New Ada compiler for .NET)

["Marin David Condic" <mcondic.auntie.spam@acm.org>]

I'm not sure what Joyce meant by "deeply embedded" - I'd guess that
differentiates from products where you run on top a commercial OS tailored
for realtime. (VxWorks, realtime flavors of Linux, etc?) I suppose there
will always be hardware out there that is not amenable to running a
commercial RTOS and so we will always need RTKs that come with compilers to
do that job. (My current computer is one such device)

I'd agree that the RTK is one of the harder parts of such a compiler, but
the good news is that it is relatively easily unbundled from the compiler
and generalized. That's kind of how we got to realtime OS's - RTKs and
commercial OS's migrating to a common point in the middle. It would seem
like a wise business decision to split off the RTK into something like RTEMS
that can be language/compiler independent and keep the compiler making OS
calls that would likely be supportable in a non-realtime environment such as
your typical workstation. (Something like the Posix calls that I believe
RTEMS supports.) That way, you've got the option of saying "I don't care if
the embedded world uses Ada or C or C++ or 'Embedded Cobol' - I can market
the RTK to them."

MDC
--
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jast.mil/

Send Replies To: m c o n d i c @ a c m . o r g

    "I'd trade it all for just a little more"
        --  Charles Montgomery Burns, [4F10]
======================================================================

Richard Riehle <richard@adaworks.com> wrote in message
news:3E1F4B6D.A8D5172F@adaworks.com...
>
> As to pricing, I would rather leave that to the respective companies.   I
can
> say
> the compilers are not inexpensive, primarily because they target embedded
> targets with carefully designed run-time environments that are intended to
> serve applications that require a high-level of dependability.   When one
> examines the world of Ada carefully, it becomes clear that creating an RTE
> is no trivial exercise.   It is easy enough to generate code, but for
embedded
> environments (what Joyce Tokar calls, "deeply embedded"),  one must
> craft the RTE so it supports all of the generated code in a predictable
way.
> It is almost like writing a small operating system for each platform.
>
> DDC-I supports a wide variety of embedded targets.    Irvine is a C-Path
> compiler so it can be ported to many different platforms.  Once again, it
> is creating that pesky RTE that makes it expensive to build such
compilers.
>
> I overlooked Green Hills in my earlier posting.  We have clients who are
> quite happy with their compilers too.
>
> The important point is that GNAT, for all of the good will toward Ada it
> has generated, is not the only game in town.   ACT has done a remarkable
> job with their current business model and produced compilers of excellent
> quality.   That is just the first step when considering "deeply embedded"
> systems.  It is not enough to generate good code.    One of the key
elements
> in selecting an Ada compiler for a "deeply embedded" system is to
> evaluate the relative qualities of the RTE.  This is a point often
overlooked
> by those who naively believe that C++ is as relevant as Ada for embedded
> systems.
>
> One reason I characterized the new owners of Aonix as a bunch of
ignoramouses
> is that they overlooked this point and laid-off some of their best people,
the
> people who knew how to differentiate their product from others by creating
> reliable, stable, and efficient RTE's.   I'm not sure whether they have
learned
> the error of their decision yet, but one can hope they have.
>
> Richard Riehle
>