Re: JNI thread synchronization

Date view Thread view Subject view Author view

From: Godmar Back (gback@cs.utah.edu)
Date: Fri Feb 12 1999 - 14:51:10 EST


 How about this:

Let's subdivide the native code in a non-JNI and a JNI part. That's already so.
Let's define that JNI is slow anyway, and non-JNI is faster.

Let's put softcalls in the JNI wrappers for the JNI part. That slows
it down some. Probably not much, though.
Let's steal a stand-alone snprintf from somewhere (probably w/o floating pt,
but maybe with).
Let's outlaw all async-signal-unsafe function in the non-JNI part and
replace them with protected functions provided the VM.
Let's have the non-JNI part use spinon/spinoff where that's too
cumbersome --- like sprintf for floating point.
If we ever get to do better gc, then just export another kaffe-specific
interface to the non-JNI part.

        - Godmar


Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:58:07 EDT