[klibc] What's next for klibc...

[H. Peter Anvin]

Olaf Hering wrote:
>  On Mon, Jul 12, H. Peter Anvin wrote:
> 
> 
>>Other than fixing the few additional architectures which have not been 
>>ported to the new syscall handling, does anyone know of anything that is 
>>needed for klibc, other than of course to update the kernel integration so 
>>that it can eventually be merged into the kernel build?
> 
> 
> Some sort of devel structure, so external stuff can link against it.
> a 'make install-devel' should install all relevant stuff to /opt/klibc or so.
> Maybe a simple shell wrapper to pass the correct options to gcc.
> dietlibc has a 'diet gcc ... ' call, no idea what it does.
> and a make install should put stuff into /lib/klibc/bin, thats what we
> use right now. scripts can then use #!/lib/klibc/bin/sh 
> 

/lib/klibc/ is pretty ugly, but /opt/klibc/ can be off-root so that's probably 
the best bet.

My inclination is to make the devel structure a Makefile skeleton, similar to 
the stuff already used in the klibc tree.  Doing something like the dietgcc 
script is of course possible, but rather complex.

	-hpa