Re: [MiNT] MiNT libs

[Martin-Eric Racine <q-funk@pp.fishpool.com>]

On Tue, 26 Jan 1999, Michael Schwingen wrote:

> On Tue, Jan 26, 1999 at 11:08:41AM +0200, Martin-Eric Racine wrote:
> > libc, blibc, portlib, libusers (and their 16 brethren) all have
> > /var/adm/wtmp embeded inside them, which I found out after doing
> > 'grep "/var/adm" *' in /usr/lib.
> 
> So I guess we should change these for a start ...

I really wonder _why_ we have utmp.c and wtmp.c in the MINTLIB,
in the first place.  shouldn't these both be integrated into ....
(can't remember exactly but something like) utmp.h or paths.h ?

Paths.h already has the paths to almost eveyrthing in there, so....

> > How does one rebuilt a lib from scratch, knowing that one needs
> > the existing lib to build the new one?
> 
> Since we do not have shared libs yet, you should not need any old lib to
> build a new one. IIRC, 'make' is all you need ...

I have tried the Makefile included with Yves MINTLIB 49, but the only
version I manage to compile (GCC 2.7.2, GAS 2.9, GCC UTIL 40) is the 
basic 68000 version _with_ -mint enabled in the Makefile:

mint.a, mint16.a, etc.

Neither of the 020 versions compile sucessfully, nor the 68000 version
_without_ -mint:

c.a, c16.a, etc.

GAS encounters an illegal instruction when it gets to the Baserel
part of the building process and GCC exits....

Question:  shouldn't the MINTLIB be programmed cleanly enough so
that it always compile without any errors, not even a signle warning?

----------------------------------------------------------------
  Martin-Eric Racine * http://www.pp.fishpool.com/~q-funk/M-E/
  The Atari TT030 Homepage * http://members.tripod.com/~TT030/
----------------------------------------------------------------