Re: NOACE- End of the road for Ada?

["Alexander E. Kopilovich" <aek@VB1162.spb.edu>]

Marin David Condic wrote:

> Q: What does Simulink generate?
> A: C
>
> You can fight it, but all the pressure from every angle will be to 
> utilize C because that's what Simulink generates.
>
> Add to that the fact that the whole model of the way these sorts of 
> tools design & operate is a kind of early vintage Fortran view of the 
> universe and even if it *did* generate Ada, you're at best going to get 
> Adatran code. How does Ada shine in any way if all you get is a stack of 
> global variables and a bunch of subroutines to operate on them? Its 
> fundamentally no better than the C you get already - so why spit into 
> the wind? You even have to fight against Ada in some ways to make it 
> work like Fortran/C so it only makes the job harder.
>
> If there were a Simulink-like thing that was modeled with Ada in mind as 
> the target language, it might have significant advantages *to the 
> Simulink modeler-guy*. Right now, the guy generating the model has no 
> reason to want to use Ada and every reason not to want to. From his 
> view, all he sees is the Simulink model and attempting to generate any 
> code in any language other than exactly what he is getting now (C) is 
> only a pain in the ass and of no particular advantage.
>
> If, OTOH, there was a Simulink-like modeling tool that took advantage of 
> Ada-ish concepts - expressed things in terms of Ada constructs, for 
> example - then the model builder would a) have the advantage of better 
> abstraction, reuse of model components, better up-front consistency 
> checks, etc., and b) would start having natural reasons to prefer Ada 
> code generation.

Well, this sounds all right, except that it raises a disturbing question:
why that thing was not done already? Or it was done, perhaps several times,
but was not published and used silently and internally only? It seems that
the thing can be really useful and really belongs to one of natural Ada
domains, so it is hard to believe that it was not done... and if it wasn't
then there should be not-too-obvious obstables or resistances. Perhaps there
are some problems with patents, licenses etc.? Or something else?

> People use CAD/CAM packages for mechanical design. Do they do any kind 
> of code generation for modeling purposes? What do they generate? Is 
> there some advantage to making the CAD/CAM tool express things in 
> Ada-ish ways? (packages? strong typing? tasking? object oriented? etc?)

I think that mechanical CAD area isn't good for Ada - mostly because of 
already established practice there. They have their "geometry engines" and
too little remains for Ada.

> What about math and statistics packages?

I don't think that math and statistics are natural domains for Ada...
especially statistics, where APL seems to be the most suitable programming
language.

> What about network applications?

I don't see any particular advantages of Ada for networking.