Re: Re(dux): Ada for small machines (was Re: ada and robots)

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

Ken said

<<I believe Jim's problems stems from the inability to build, using Ada,
"REAL CONTROL apps running in 89C1051's, with 1K of FLASH ROM." This
is, of course, due to the lack of Ada compilers in this area, possibly
caused in part by the need to build a tasking RTE for this target, and
most definitely caused in part by the perception by Ada vendors that
no one would use Ada for such a target. (This perception did not stop
Jim's company from building a Modula compiler for this target, however).
>>

The issue of building the tasking runtime is a minor one. The issue of
not believing that there would be a market is the real issue. Speaking
for us we build ports if any of the following is true

  a) it is clear that there is a wide demand for the port
       (demand = demand from paying customers)

  b) the port is funded (that is why there is an OS/2 port for example,
	the government funded this port)

  c) it is easy to build the port, and it seems useful enough to be worth
	the effort (the Linux port got built this way initially)

  d) some volunteer contributes the port (the DOS port got built this way)

None of these factors seems to be present for the 89C1051.

Sure, if there is no Ada compiler for a machine, the best technical advice
is not to consider writing in Ada. But what happens in this and many similar
threads is to make the jump from "there is no Ada compiler" to "It is 
impossible or difficult to build an Ada compiler" for a given target.

I am sure that Jim's company uses a similar list to the above, and I can
think of any number of reeasons why the calculation looks different for
him than it does to us.