Re: Garbage Collection in Ada

[jsa@alexandria (Jon S Anthony)]

In article <dewar.845386292@merv> dewar@merv.cs.nyu.edu (Robert Dewar) writes:

> Many people claim many things. One thing to be careful of is that people
> often justify not using Ada because Ada lacks xxx, and then rush off and
> use some other language that lacks xxx.

Yes, that's certainly true.  But of the cases I have in mind that is
not true.


> In the case of GC, there are not many languages that in practice compete
> with Ada that supply GC, so I doubt these claims. If and when Java becomes
> a pratical vehicle for the kind of applications that might otherwise be
> written in Ada, this situation may change, but I suspect the claim above
> is valid.

Depends on what you mean by "compete".  Does Eiffel?  Does SmallTalk?
Probably.  But they are not the big market either (though SmallTalk
may be getting there...)  OTOH, if you wait for Java to take the lead,
then there is no reason to believe that any good will come of this for
Ada users.


> In particular, our experience is often that people making such claims don't
> even want to pay for *any* support, what they want is a version of GNAT they
> can use free that has the feature they want. That's a perfectly reasonable
> thing to want, but wanting does not make it so!

You're not actually suggesting that I'm not paying, are you????!!  I
can give you my customer # if you are...


> Clearly any Ada company will fund developments that it thinks might 
> interest potential customers. So far, we don't buy the argument that
> investing $x in implementing GC will increase our support income by
> $x (not to mention more than $x), so it's not on our list of priorities.

Fair enough.  But then, I find it hard to believe that implmementing
all the annexes had that level of payback either.  Except for the fact
that you can claim to fully support the entire RM (another good reason
why GC annex may have been nice :-).  Think about it, for example, the
DSA is a fine and wonderful thing but I can get everything it offers
in commercially available CORBA based ORBs supporting Ada95.  Not so a
GC...)


> You have to remember here that I am an enthusiastic supporter of the notion
> of garbage collection (see for example my paper in 1977 on an interesting
> new garbage collection algorithm used in the Macro-SPITBOL compilers in
> SOftware Practice and Experience). 

Yes, I believe this entirely.


> I am a little dubious about mixing GC with full Ada 95 (e.g. GC and
> unchecked conversion between integers and pointers are not friends),

Agreed, but these sorts of things could be requirements on when GC is
available.  Just like the various requirements for other special needs
aspects.


> but certainly GC with a slightly restricted Ada 95 semantics is perfectly
> practical, and you can add me to the people who would like to see it (but
> are not willing to sign a big check for it!)

I hear ya!  I wonder if there is enough _small_ folk out there who
_together_ could fund such an effort?  What sort of $$ are we talking
about?

/Jon
-- 
Jon Anthony
Organon Motives, Inc.
Belmont, MA 02178
617.484.3383
jsa@organon.com