[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] osbind.S ?



Frank Naumann wrote:
I see the #define Fseek64 but I can't find the #define for trap_1_wllwwl.

Sorry, I made a stupid mistake: I looked for "trap_1_wllw" (the one with invalid stack) with partial word matching, so I found "trap_1_wllwwl" too, which is used by Fseek64(). And you're right, there is a second different problem here, the definition if "trap_1_wllwwl" is missing.

I remember Alan has made some work around 64-bit wrappers (not real 64-bit support), he will tell us what to do with this invalid Fseek64() macro.

But first I add the return thing to the syscalls.master. I think, it's neccessary (not only for Fgetdta()).

Yes, it would be better to have an exact compatibility with current implementation.

The generator is a good thing, it will be soon ready to replace osbind.h, mintbind.h, falcon.h and metados.h. However, it would probably be good to split syscalls.master into osbind.master, mintbind.master, and so on. So the different "libraries" will be kept separate. It should be possible to put arbitrary text into the *.master files, to be copied verbatim to the generated files, for example for comments, defines and structures. That way, it will be a perfect replacement for current hand-written headers.

We must not forget the Line A. linea.h is similar to osbind.h. I can't see the interest of linea.c (functions instead of macros), it should be removed. The Line A should be treated in the same way as other system calls.

So things are becoming better :-)

--
Vincent Rivière