Re: CVS commit: pkgsrc/pkgtools/pkgin

[David Holland <dholland-pkgchanges%netbsd.org@localhost>]

On Mon, Oct 17, 2011 at 11:44:12AM +0200, iMil wrote:
 > >Actually I am able to reproduce this behaviour on 5.1 with current sqlite3
 > >by setting WARNS >= 3.
 > 
 > And as a matter of fact, i see that
 > src/external/public-domain/sqlite/Makefile.inc has WARNS=2, just
 > like i do in pkgin's Makefile. Do you by any chance override WARNS
 > value ?

Nowhere that I can see.

I thought that the default WARNS had been raised a while back, but I
can't find any evidence of this in--...

Oh. I see what's going on, and this probably explains many of the
random divergences in compiler behavior Joerg and I have been seeing
too: my bulk build is bootstrapped, because it uses nonstandard paths,
so it's using the *.mk files that get provided that way. Those have
only two levels of WARNS:

.if defined(WARNS)
.if ${WARNS} > 0
CFLAGS+= -Wall -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith
# XXX Delete -Wuninitialized by default for now -- the compiler doesn't
# XXX always get it right.
CFLAGS+= -Wno-uninitialized
.endif
.if ${WARNS} > 1
CFLAGS+=-Wreturn-type -Wcast-qual -Wpointer-arith -Wwrite-strings
CFLAGS+=-Wswitch -Wshadow
.endif
.endif

It's possible that this should be updated in pkgsrc to match NetBSD
base to avoid unnecessary confusion.

-- 
David A. Holland
dholland%netbsd.org@localhost