Hi Torsten, Mark Wielaard wrote:
>Hi Torsten, > >On Thu, 2004-08-19 at 10:45 +0200, Dr. Torsten Rupp wrote: > > >>I'm currently changing the target native layer to replace the macros by >>functions. >> >> > >Please take a look at the NATIVE-LAYER branch that Guilhem is working >on. He already replaced all macros by functions. He is using the JNI >types as datatypes. See also the discussion in this thread: >http://thread.gmane.org/gmane.comp.java.classpath.devel/6695 >And the various patches marked [NATIVE-LAYER] in classpath-patches. > >Cheers, > >Mark > > I am following Mark's remark. I have completely removed target layer in the NATIVE-LAYER branch. As we are not limited by the strong assumptions you need on the compiler (not necessarily 64 bits, etc...) I have taken some liberties on the implementation. However it has other advantages: as being really thread safe, portable to other thread system and using functions. Maybe we could inline everything if really needed but I doubt we really lose time/cpu cycles in just one more direct function call in the JNI implementation. The default implementation should follow posix approximately (there are surely room for enhancements). I used everywhere JNI types. However it is not yet finished because there are still syscalls lying here and there in JNI function calls (they have been added independently from the TARGET layer). So, please take a look at it and tell me of any remark. Regards, Guilhem.
