Ada subset -- why not?

Ada subset -- why not?

[Jim Vijay]

Given:
 1. The vast majority of programming in the real world is non-real-time.
 2. Compilers with full-blown real-time support are more expensive - in money,
    and user resources.
 3. Ada (83 & 95) is cleanly separable syntactically into non-real-time and
    real-time subsets.
 4. A non-real-time subset would introduce most of the software engineering
    concepts/benefits of Ada and thus promote the usage of the real Ada.

If these statements are true, why is there not an Ada-lite on the market?
Does the Ada JPO have specific reasons to prevent such a subset language?
Can they legally prevent someone from marketing such a language?
Has anyone tried?  One would think it would be possible for Borland (say)
to upgrade their Object Pascal to Ada-lite 95.  Think of the marketing
edge they would have trumpeting its OO, safety, reliability, LOW PRICE,
GREAT PERFORMANCE, AND UPWARDS COMPATIBILITY etc.  If it takes off, they
could later add the real-time portion, possibly as an option.

I know that a free GNAT exists.  But a market driven solution would be
more attractive - even if its (initially) a lite solution.

-- Jim Vijay

This message is mine - that is to say it belongs to me. (only)

Re: Ada subset -- why not?

[Michel Gauthier]

In article <vijay-0102951423400001@vijay.chinalake.navy.mil>,
vijay@nawc690.chinalake.navy.mil (Jim Vijay) wrote:

>>  [...] why is there not an Ada-lite on the market? [...] 

It is a very old dabate.
My experience is that we should take great care about Ada subsets, even
in the case of teaching uses. Let me argue.

Of course, students do not build complex programs, which leads some
lecturers to think that an Ada subset is sufficient for them.
My experience is that some basic libraries should be given to the students,
so that they can experiment not-invented-here components very early
in their life. To be used in this frame, these components should be very
robust (exceptions, messages, exotic cases), and I could never reach
the expected level of robustness with an Ada subset.

Of course, real-time features and representation clauses might be
avoided, but all of genericity, exceptions, overloading, tasking, VERY
strong typing have to be kept in the basic language.

-- 
Michel Gauthier - Laboratoire d'informatique - 123 avenue Albert Thomas
F-87060 Limoges - fax +33()55457315
----- Are the messages that objects exchange also objects ? -----

Re: Ada subset -- why not?

[Jim Vijay]

Before a lot of people reply to correct me, let me do so.  By "real-time"
I mean multitasking and protected objects.  (Yes I know that is not an
exact definition.  But I don't want an net war to try and decide what
real-time means either.)  Ada lite would cantain no support for these. 
Sorry about the omission.

This message is mine - that is to say it belongs to me. (only)