Pierre d'Herbemont wrote:
Le 25 août 04, à 23:01, Jim White a écrit :
I would really like to get away from having to write all the little wrappers to deal with the namespace problem. I believe we can solve that with a suitable naming convention as we began to do with the winegcc patch.
I think we can forget this solution, loading symbol at launch time works like a charm :)
(BTW it was indeed winebuild which was patched)
Pierre.
While I think that dynamic binding can work, I'm concerned about calling conventions between X86 and PPC. We'll need to look into what those are and what QEMU uses. It may turn out that to get things to line up correctly and not have anything get clobbered, we might not be able to allow any intermediate calls. In which case we have to use an emulation escape so that the call can be translated as a direct native call rather than using a syscall. And in that case the indirection through the pointer is likely be a problem.
Besides not having little wrappers for the namespace issue is not having to write syscalls for all the OS X API we want to use. I'ld much rather folks could just call the OS X API directly (albeit with carefully written code). Of course the advantage to going this (syscall) route is that it is easy to write and doesn't depend on esoteric compiler/linker/loader/emulator coordination.
Of course if you're planning on compiling Wine as PPC then we're on different pages architecturally.
Jim