From: | mbk@caffeine.engr.utk.edu (Matt Kennel) |
Newsgroups: | comp.lang.java,comp.lang.c++,comp.compilers |
Followup-To: | comp.lang.java,comp.compilers |
Date: | 10 May 1996 01:31:16 -0400 |
Organization: | University of Tennessee, Knoxville and Oak Ridge National Laboratory |
References: | 96-05-036 96-05-064 |
Keywords: | Ada, Java |
Stephen A. Leake (sal714@rs710.gsfc.nasa.gov) wrote:
: Ada95 has been ported to the JVM by Intermetrics in their Applet Magic
: compiler (I believe this is only in beta release now). See
: http://www.inmet.com/ada.html. Since Ada supports pointers, I assume
: the JVM does, at least in the safe way that Ada allows!
Ada has 'access types' which are like 'references' in Java and Eiffel,
but no natural pointer arithmetic.
The JVM is much more compatible with Ada95 and Eiffel than C++.
Did anybody see the Sunday New York Times? There was an article about
economists' research on "path dependence": how some choices
(especially in high tech) can prevent the system from reaching global
optimum?
For example, locking in to inferior technology at the outset
(e.g. MSDOS) can suppress development and acceptance of something
better for a long time.
Apparently this is shocking news to economists, though plainly obvious
to physical scientists: local optimization in a nonlinear system does
not lead to global optimum.
I think you can see the implications here.
I.e: do NOT pervert the JVM to C (and C++) semantics. That would mean
the prolonged continuation of inferior application technology and languages.
It might be somewhat painful at first (to have no easy C/C++ -> JVM
converter) but I'm sure that in the long run this will be the right
decision. No doubt a C++ converter is "possible" but if it means a 2x
performance hit compared to java or Eiffel or Ada...
We should adapt the JVM to the future (or at least the present capability
of Dylan/CLOS/Smalltalk/Eiffel types of languages).
--
Return to the
comp.compilers page.
Search the
comp.compilers archives again.