Kiyo Inaba wrote:
Dalibor,
My earlier patch I posted is just to verify where we have problems. I am still thinking whether macro version is better or switch to inlining.
Any idea?
I'd rather have it inlined, since inlining implies better type checking from the compiler, more meaningful warnings, and is in general more useful in conjuction with a debugger, for example. :)
Since the problem seems to be the dynamically allocated array, how about using __bultin_alloca from gcc to allocate it, like we do on powerpc, ia64 and x86_64 ? At least on powerpc-darwin, Riccardo's logs from today [1] show no warnings from the compiler wrt to inlining sysdepCallMethod.
If you agree with that, I could write up a patch to switch it over to use __builtin_alloca.
cheers, dalibor topic
[1] http://multix.dyndns.org/Sites/kaffe-devel/machines/atrejus/Atrejus.html
_______________________________________________ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
