On Sat, Jun 19, 2004 at 01:40:30PM +0200, Hubert Feyrer wrote:
> 
> One one of my machines (not the other!) I get this, trying to compile 2.0
> sources on 1.6.2 userland:
> 
> 	cc -O -I. -I./include -I/usr/cvs/src-current/tools/compat -DHAVE_NBTOOL_CONFIG_H=1 -D_FILE_OFFSET_BITS=64 -D_NETBSD_SOURCE -D__DBINTERFACE_PRIVATE -c -o pw_scan.lo.o /disk4/cvs/src-current/tools/compat/../../lib/libc/gen/pw_scan.c
> 	/disk4/cvs/src-current/tools/compat/../../lib/libc/gen/pw_scan.c: In function `__nbcompat_pw_scan':
> 	/disk4/cvs/src-current/tools/compat/../../lib/libc/gen/pw_scan.c:96: `_PASSWORD_NOWARN' undeclared (first use in this function)
> 	/disk4/cvs/src-current/tools/compat/../../lib/libc/gen/pw_scan.c:96: (Each undeclared identifier is reported only once
> 	/disk4/cvs/src-current/tools/compat/../../lib/libc/gen/pw_scan.c:96: for each function it appears in.)
> 	...
> 
> _PASSWORD_NOWARN etc. are defined only if _NETBSD_SOURCE is set, which
> should be the case given that it's set on the command line.

Yes...there are a number of problems here (and some PR's filed even for it).

I'm in the process of purging all uses of _NETBSD_SOURCE from the tools/compat
builds and providing compat defs for anything which falls through the cracks.

Just testing is taking a while to complete (2.0 on x86, amd64, 1.6.2, linux,
freebsd and a solaris host). So far I'm up to clean builds on linux and 2.0
and 1.6.2 should be going shortly. Once I regress freebsd/solaris I'll commit
(hopefully tonight)

> 
> I don't understand the maze of the build process enough to debug this
> further... what am I doing wrong?

it is a maze today. Hopefully some of this will also make it simpler to
deal with.

James