Re: libjava status on Solaris 8/Intel and IRIX 6.5

[Andrew Haley <aph at cambridge dot redhat dot com>]

Tom Tromey writes:
 > >>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:
 > 
 > Rainer> Indeed.  Same on sparc-sun-solaris2.8, so this is a generic
 > Rainer> Solaris 2 problem.  We probably need a new spec substitution
 > Rainer> for this (similar to DIVIDESPEC and friends)?  Any suggestions
 > Rainer> for a good name, and I'll provide a patch and verify it.
 > 
 > It doesn't matter too much.  REFERENCESPEC or something like that
 > would be fine.
 > 
 > I assume you're going to check in your Solaris DIVIDESPEC change?

I don't have a machine to test on.

 > Andrew> The trouble with -fcheck-references is that it only works for
 > Andrew> Java code and not in libgcj's CNI C++ code, so using it for
 > Andrew> tests may give a false sense of security.  Perhaps we should
 > Andrew> not do this in order not to mislead users.
 > 
 > Our own CNI code probably assumes that such dereferences will be
 > caught at runtime :-(

Ah, probably.

 > Andrew> Hey Tom, how about redesigning CNI to use smart pointers?  :-)
 > Andrew> (Actually only half a smiley; it might be worth doing for CNI 2.0...)
 > 
 > We had a long discussion about this once.  It is in the archives
 > somewhere.  There are some real disadvantages, namely massive
 > uglification of the code.

It's not ugly, just different.  Instead of

   void barf (java::lang::grope *poo)

you have

   void barf (ref<java::lang::grope> poo)

and carry on as usual.

Andrew.