Le 07/11/2010 10:17, assembling signals a écrit : > Hello again! > > > For that compiler error which you could reproduce, will you file a bug report, > or is the API that much under change currently, that it's not relevant yet? > Sorry, the proposed workaround produces same error for me. >
Works with javac. Do you use it ? > For the use case "calling a method whose signature is unknown prior to > runtime" > (such as some lib, loaded externally) : > - Is invokeVarargs the only method which could "replace" the old slow > Method.invoke for the given use case? > - As mentioned in first mail, my benchmark showed for invokeVarargs > nearly same slowness as Method.invoke. > - So, the given use case doesn't profit from new concept at all? > In that case: :'( > Why do you want to call a method with a signature which is unknown prior runtime ? You want to write an interpreter ? In that case, you can generate bytecode at runtime ane becuase you do that at runtime you know at least the number of parameter. > > Best regards, > Ivan G S > cheers, Rémi _______________________________________________ mlvm-dev mailing list mlvm-dev@openjdk.java.net http://mail.openjdk.java.net/mailman/listinfo/mlvm-dev
