Re: Ada GC (was about Java VM)

"For your information, we are working on Java virtual machine support
that will provide garbage collection for hard-real-time embedded
applications. See the web site mentioned in my signature for more
details."


Sounds perfectly reasonable. As Henry said, the basic problems here
have been solved at a conceptual level a long time ago. One of Bill
Van Der Poel's students wrote a PhD thesis in the early 70's that was
one approach to a complete solution of the RT/GC problem (I think this
thesis was only ever published in Dutch), but there are many other
papers from this period.


But this does not mean that RT/GC is well enough established
technology to be widely used and required yet.


After all, in 1980 when Ada was designed, object oriented programming
was very well understood (remember that Simula 67 was nearly 15 years
old at that stage), and furthermore, the designers of Ada 83 carefully
considered whether to incorporate OO design aspects in Ada 83, and
decided not to (they had a lot of experience with Simula). They
decided that it was premature, and in retrospect that was probably
right (we learned a lot about the best way of doing OO in the decade
that followed).


I suspect we are at a similar stage with RT/GC. The technology begins
to be there, but I doubt it is mature enough to incorporate into
mainstream language design yet. As I mentioned, I at least tried to
get GC into Ada 95 as part of the IS annexe, but even there it seemed
over-reaching and I found abslutely NO support for this position.


It is indeed interesting that JAVA assumes GC, and this will spark
some more interesting work in this area.
--

Return to the comp.compilers page.
Search the comp.compilers archives again.