Re: Array classes and class loaders

[Andrew Haley <aph at redhat dot com>]

Tom Tromey writes:
 > Bryce> Okay, I RTFM'ed and I think I understand this now. Problem is
 > Bryce> the "initiating class loader" is always going to be null until
 > Bryce> the compiler is modified to pass the calling class to things
 > Bryce> like _Jv_NewArray. Oh well.
 > 
 > I think the current plan is to instead somehow modify the runtime to
 > be able to find the caller's class, perhaps by the same kind of
 > mechanism we use to generate the stack trace.  Maybe we'd have to add
 > code to the compiler for platforms where we can't do the backtracing
 > trick any other way.

Perhaps, but I can't help thinking the time would be better spent
making backtrace() work.  I'll grant you this is harder on some
platforms than others, but eventually it even worked on the IA-64.
Anything after that is easy.  :-)

 > (Also there's the issue of finding the class for interpreter
 > frames.)

I've been thinking about that too.  It's an interesting one, but we
need to be able to do this.  Trampolines might work...

Andrew.