Re: pkg/48412 (sysutils/cdrtools fails to build on NetBSD-6.1.2/amd64)

[Greg Oster <oster%cs.usask.ca@localhost>]

The following reply was made to PR pkg/48412; it has been noted by GNATS.

From: Greg Oster <oster%cs.usask.ca@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: pkg/48412 (sysutils/cdrtools fails to build on
 NetBSD-6.1.2/amd64)
Date: Thu, 28 Nov 2013 13:42:32 -0600

 On Thu, 28 Nov 2013 17:35:00 +0000 (UTC)
 Greg Oster <oster%cs.usask.ca@localhost> wrote:
 
 > The following reply was made to PR pkg/48412; it has been noted by
 > GNATS.
 > 
 > From: Greg Oster <oster%cs.usask.ca@localhost>
 > To: gnats-bugs%NetBSD.org@localhost
 > Cc: 
 > Subject: Re: pkg/48412 (sysutils/cdrtools fails to build on
 >  NetBSD-6.1.2/amd64)
 > Date: Thu, 28 Nov 2013 11:28:41 -0600
 > 
 >  On Thu, 28 Nov 2013 16:45:00 +0000 (UTC)
 >  Matthias Scheler <tron%zhadum.org.uk@localhost> wrote:
 >  
 >  > The following reply was made to PR pkg/48412; it has been noted by
 >  > GNATS.
 >  > 
 >  > From: Matthias Scheler <tron%zhadum.org.uk@localhost>
 >  > To: oster%netbsd.org@localhost
 >  > Cc: gnats-bugs%NetBSD.org@localhost
 >  > Subject: Re: pkg/48412 (sysutils/cdrtools fails to build on
 >  >  NetBSD-6.1.2/amd64)
 >  > Date: Thu, 28 Nov 2013 16:42:09 +0000
 >  > 
 >  >  On Thu, Nov 28, 2013 at 03:35:00PM +0000, Greg Oster wrote:
 >  >  >  >  Yeah, it's hidden specifically to keep configure tests from
 >  >  >  > finding and latching onto that stuff.
 >  >  >  
 >  >  >  The test for openat() seems to be just the equivalent to:
 >  >  >  
 >  >  >   #include <fcntl.h>
 >  >  >   int main()
 >  >  >   {
 >  >  >     openat();
 >  >  >   }
 >  >  >  
 >  >  >  which compiles just fine on NetBSD 6.1.2.  That compilation
 >  >  > enables HAVE_OPENAT, which turns off the schily version, but is
 >  >  > not sufficient to guarantee the availability of the AT_FDCWD
 >  >  > #defines from /usr/include/sys/fcntl.h :(
 >  >  
 >  >  Correct. This is why my patch tries to bypass the "configure"
 >  > script and deactivates all the configure checks for these
 >  > function. But I don't understand why this doesn't work on your
 >  > system.
 >  >
 >  >  Are you perhaps setting "PKGSRC_SETENV" to a non-default value?
 >  
 >  Nope... that variable is not set.
 >  
 >  Something is certainly fishy, as it fails to build on 'mickey', but
 >  builds fine on 'quad', and those are both 6.1.2/amd64 boxes... 
 >  
 >  On 'mickey' I see:
 >  mickey# grep -r ac_cv_func_openat *
 >  
 > work.mickey/cdrtools-3.01/incs/amd64-netbsd-gcc/config.cache:ac_cv_func_openat=${ac_cv_func_openat=no}
 >  
 > work.mickey/cdrtools-3.01/incs/x86_64-netbsd-gcc/config.cache:ac_cv_func_openat=${ac_cv_func_openat=yes}
 >  
 > work.mickey/cdrtools-3.01/cdda2wav/OBJ/amd64-netbsd-gcc/config.cache:ac_cv_func_openat=${ac_cv_func_openat=no}
 >  
 >  
 >  but on 'quad' I see:
 >  quad# grep -r ac_cv_func_openat *
 >  
 > work.quad/cdrtools-3.01/incs/x86_64-netbsd-gcc/config.cache:ac_cv_func_openat=${ac_cv_func_openat=no}
 >  
 > work.quad/cdrtools-3.01/cdda2wav/OBJ/x86_64-netbsd-gcc/config.cache:ac_cv_func_openat=${ac_cv_func_openat=no}
 >  
 >  Why would mickey have amd64-netbsd-gcc and not quad?  And then why
 >  wouldn't it get x86_64-netbsd-gcc correct!?!?
 >  
 >  Gotta be something in my configs on these machines... (though not
 > sure what... and both are building in chroots... )
 >  
 >  Hmm.. quad does:
 >  
 >          ==> MAKING DIRECTORY "incs/x86_64-netbsd-gcc/Inull"
 >          ==> CONFIGURING RULES "incs/x86_64-netbsd-gcc/rules.cnf"
 >  creating cache ./config.cache
 >  checking host system type... amd64-unknown-netbsd6.1.2
 >  checking if sh is bash... no
 >  checking if /bin/sh is bash... no
 >  checking whether sh -ce is broken... no
 >  checking whether /bin/sh -ce is broken... no
 >  checking whether /bin/bosh is a working shell... no
 >  checking whether /opt/schily/bin/bosh is a working shell... no
 >  checking for cc... gcc
 >  checking for gcc... (cached) gcc
 >  checking whether the C compiler (gcc   -O   -Llibs/x86_64-netbsd-gcc
 >          -L/opt/schily/lib) works... yes checking whether the C
 > compiler (gcc   -O   -Llibs/x86_64-netbsd-gcc -L/opt/schily/lib) is a
 >          cross-compiler... no checking whether we are using GNU C...
 > yes checking whether gcc accepts -g... yes
 >  
 >  while mickey does:
 >  
 >          ==> MAKING DIRECTORY "incs/amd64-netbsd-gcc/Inull"
 >          ==> CONFIGURING RULES "incs/amd64-netbsd-gcc/rules.cnf"
 >  creating cache ./config.cache
 >  checking host system type... amd64-unknown-netbsd6.1.2
 >  checking if sh is bash... no
 >  checking if /bin/sh is bash... no
 >  checking whether sh -ce is broken... no
 >  checking whether /bin/sh -ce is broken... no
 >  checking whether /bin/bosh is a working shell... no
 >  checking whether /opt/schily/bin/bosh is a working shell... no
 >  checking for cc... gcc
 >  checking for gcc... (cached) gcc
 >  checking whether the C compiler (gcc   -O   -Llibs/amd64-netbsd-gcc
 >          -L/opt/schily/lib) works... yes checking whether the C
 > compiler (gcc   -O   -Llibs/amd64-netbsd-gcc -L/opt/schily/lib) is a
 >          cross-compiler... no checking whether we are using GNU C...
 > yes checking whether gcc accepts -g... yes
 >  
 >  hmmmmm....
 
 Long story short:
 
 I had 'ARCH' defined to 'amd64' in my shell environment on mickey.....
 that was causing conf/oarch.sh to spit out 'amd64' instead of
 calculating 'x86_64', and blowing things up.
 
 *sigh*
 
 Later...
 
 Greg Oster