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

Re: [MiNT] Mintlib



Hi Mark,

I wonder why would you want to quit using mintlib when porting apps to be usable under other systems. Absolute mintlib functions has fallbacks when the MiNT syscall is not present which goes down to more available functions. So what is the problem there? If you need networking to work under different OS then you can simply try to create the sys/socket.h wrappers and implement them different way for MiNT and the OS you'd like to support and link this library ahead mintlib so that your wrappers are the ones linked into.

MagiXNet? How many people succeeded to use this? Hmm, the 1.16.x kernels natively provide syscalls that are BSD socket compatible (GEMDOS extensions 0x160-0x16d) which are used from mintlib. If I were a MagiC supporter (I was at some point as I didn't have powerfull machine to run MiNT on that time) I would go for implementing those syscalls. AFAIK MagiXNet is actually 1.15.x mintnet port where the /dev/socket is the the implemented thing as all the sys/socket.h calls were done though Fcntl() there.

best regards

STanda


Mark Duckworth wrote:
Hey guys,

Lately I've been experimenting with getting GIM running under other OS's
than MiNT.  It seems the most recent standalone mintlib that could
compile under other compilers like purec is 048.  I got this compiled
and I tried to start subtituting libraries and crt0, and gcrt0 but I
have hit another snag.  Missing functions.  Certain things that libfaim
needs like snprintf aren't present.  I could try to backport those
functions into the standalone mintlib but I decided instead to go
another route.

Is there documentation anywhere of what functions in mintlib explicitly
depend on the kernel now when they didn't always before?  Are we
modeling a linux approach by building some of this stuff into the kernel
or?  I'm unsure of the motivation for this, except perhaps this a way
around our static only problem.  Basically is it still possible to port
unix apps to MagiC/Magxnet or am I simply barking up the wrong tree?
It'd be awfully nice of the magxnet code was released with sourcecode
and methodology but I guess that was too much to ask.

Thanks,