Re: newbie Q: storage management

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

Jon said

<<False.  You can pretty much get something like this now using a bolt
on fully conservative collector.  As you note, portability between
compilers is a significant part of the problem.  As I've noted before,
this alone would make the above scenario irrelevant for us (and we're
someone who really wants GC!)  But that is not all.  There are support
issues, synchronization issues with GNAT releases, platform
portability issues, etc.  At the moment, wrt GNAT, the only plausible
way for these issues to be sorted out would be if the "someone"
mentioned above turned out to be ACT.>>

Not at all, a lot of useful things have been added to GNAT by people
working outside ACT. After all remember the one and only reason that
the DoD funded the GNAT activity was to provide a research vehicle
for such development efforts (the funding was not at all interested
in producing a student usable compiler, or in a generally usable
compiler at all -- the fact that we achieved more was because of
our determination to solve a wider range of problems!)

Obviously the only way ACT would get interested in doing work on GC, like
any other vendor, is if there were significant customer demand for such
work. Right now, there is no customer demand whatsoever, so the likelihood
of us doing any work on GC is zero.