Re: liblua (CVS commit: src/distrib/sets/lists)

[Alistair Crooks <agc%pkgsrc.org@localhost>]

On Sun, Oct 31, 2010 at 11:49:33AM -0700, Paul Goyette wrote:
> I notice that for libm, libmagic and others, base/md.amd64 contains only
> two entries
>       lib/i386/libxxx.so.x
>       lib/i386/libxxx.so.x.y
> 
> while there are five entries in comp/md.amd64
>       lib/i386/libxxx.a
>       lib/i386/libxxx.so
>       lib/i386/libxxx_p.a
>       lib/i386/libxxx_pic.a
>       lib/i386/libxxx_g.a
> 
> Yet, this commit places lib/i386/liblua.so in base/md.amd64 list, and 
> there does not seem to be a lib/i386/liblua_g.a
> 
> I'm not trying to complain about the current commit,but only trying to 
> determine if we should be consistent!  Is there a valid reason why 
> lib/i386/liblua.so belongs in base/md.amd64 while lib/i386/libm.so goes 
> in comp/md.amd64 ?

My gensetlist stuff (referenced on tech-userlevel, available here)

        http://www.netbsd.org/~agc/gensetlist-20101031.tar.gz

puts the compat shared objects in base/md.amd64 (and
base/ad.mips64{eb,el}, and base/md.sparc64), and the rest of the
compat entries in

[src/distrib/sets/lists/comp/ad.mips64eb]
./usr/lib/64/libnetpgp.a                comp-c-lib      compat,crypto
./usr/lib/64/libnetpgp.so               comp-sys-shlib  compat,crypto
./usr/lib/64/libnetpgp_g.a              comp-c-proflib  compat,debuglib,crypto
./usr/lib/64/libnetpgp_p.a              comp-c-proflib  compat,profile,crypto
./usr/lib/64/libnetpgp_pic.a            comp-c-piclib   compat,pic,crypto
./usr/libdata/debug/usr/lib/64/libnetpgp.so.3.0         comp-compat-shlib       
compat,pic,debug,crypto

so we may well be missing the debug entry for the compat shlib?

Please note that this was done with a contrived example of a modified
netpgp Makefile, not lua, but I think the principles still apply.

Regards,
Alistair