I mentioned this on IRC but Ville said that ABCL methods where much smaller and would not be an issue.
So based of this info I made I never inline bytecode in larkc/uabcl .. (I make it poke at well heated static code instead preferred.. this also makes the system more portable crossed recompilations). when a static call site is not prefered a non static but well visited call site works well ex: foo.length() if CXRImpl.LENGTH(LispObject obj) is too annoying in bytecode. so foo.length() is the reasonable compromise
On Thu, Jun 17, 2010 at 4:43 AM, Mark Evenson evenson@panix.com wrote:
On 6/8/10 1:41 PM, Mark Evenson wrote:
http://nerds-central.blogspot.com/2009/09/tuning-jvm-for-unusual-uses-have-s...
The point of posting this: if this article is correct, no methods over 8k are compiled by the JIT. This should be verified.
-- "A screaming comes across the sky. It has happened before, but there is nothing to compare to it now."
armedbear-devel mailing list armedbear-devel@common-lisp.net http://common-lisp.net/cgi-bin/mailman/listinfo/armedbear-devel