Re: pkg/36441: www/thttpd build fails in current pkgsrc

[Aleksej Saushev <asau%hotbox.ru@localhost>]

The following reply was made to PR pkg/36441; it has been noted by GNATS.

From: Aleksej Saushev <asau%hotbox.ru@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: pkg-manager%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
        pkgsrc-bugs%netbsd.org@localhost
Subject: Re: pkg/36441: www/thttpd build fails in current pkgsrc
Date: Tue, 05 Jun 2007 23:19:36 +0400

 "OBATA Akio" <obache%netbsd.org@localhost> writes:
 
 > Build problem isn't happened on my environment(NetBSD 3.1 and pkgsrc-HEAD).
 > What is occured on your environment? Please present a error message.
 
 I use netbsd-HEAD and pkgsrc-HEAD.
 
 To be more informative, I ran it with 'make MAKE_JOBS=1 >build.log 2>&1'
 since I have 'MAKE_JOBS?=2' in my /etc/mk.conf, I get this, in short:
 
 creating cache ./config.cache
 checking host system type... i386--netbsdelf
 checking target system type... i386--netbsdelf
 checking build system type... i386--netbsdelf
 
    [Lots of configuration]
 
 checking for a BSD compatible install... /usr/bin/install -c -o root -g wheel
 updating cache ./config.cache
 creating ./config.status
 creating Makefile
 creating cgi-src/Makefile
 creating extras/Makefile
 ===> Building for thttpd-2.25bnb5
 for i in cgi-src extras ; do (  cd $i ;  pwd ;  /usr/bin/make   -j 1  
WEBDIR=/usr/pkg/share/thttpd  CGIBINDIR=/usr/pkg/libexec/cgi-bin  
MANDIR=/usr/pkg/man  WEBGROUP=www  ) ; done
 /usr/pkgsrc/www/thttpd/work/thttpd-2.25b/cgi-src
 cc -O2 -O3 -march=i686 -O2 -DHAVE__PROGNAME=1 -DHAVE_FCNTL_H=1 -DHAVE_GRP_H=1 
-DHAVE_MEMORY_H=1 -DHAVE_PATHS_H=1 -DHAVE_POLL_H=1 -DHAVE_SYS_POLL_H=1 
-DHAVE_SYS_EVENT_H=1 -DTIME_WITH_SYS_TIME=1 -DHAVE_DIRENT_H=1 -DHAVE_LIBCRYPT=1 
-DHAVE_STRERROR=1 -DHAVE_WAITPID=1 -DHAVE_VSNPRINTF=1 -DHAVE_DAEMON=1 
-DHAVE_SETSID=1 -DHAVE_SETLOGIN=1 -DHAVE_GETADDRINFO=1 -DHAVE_GETNAMEINFO=1 
-DHAVE_GAI_STRERROR=1 -DHAVE_KQUEUE=1 -DHAVE_SIGSET=1 -DHAVE_ATOLL=1 
-DHAVE_UNISTD_H=1 -DHAVE_GETPAGESIZE=1 -DHAVE_MMAP=1 -DHAVE_SELECT=1 
-DHAVE_POLL=1 -DHAVE_TM_GMTOFF=1 -DHAVE_INT64T=1 -DHAVE_SOCKLENT=1 -I.. -c 
redirect.c
 cc -O2 -O3 -march=i686 -O2 -DHAVE__PROGNAME=1 -DHAVE_FCNTL_H=1 -DHAVE_GRP_H=1 
-DHAVE_MEMORY_H=1 -DHAVE_PATHS_H=1 -DHAVE_POLL_H=1 -DHAVE_SYS_POLL_H=1 
-DHAVE_SYS_EVENT_H=1 -DTIME_WITH_SYS_TIME=1 -DHAVE_DIRENT_H=1 -DHAVE_LIBCRYPT=1 
-DHAVE_STRERROR=1 -DHAVE_WAITPID=1 -DHAVE_VSNPRINTF=1 -DHAVE_DAEMON=1 
-DHAVE_SETSID=1 -DHAVE_SETLOGIN=1 -DHAVE_GETADDRINFO=1 -DHAVE_GETNAMEINFO=1 
-DHAVE_GAI_STRERROR=1 -DHAVE_KQUEUE=1 -DHAVE_SIGSET=1 -DHAVE_ATOLL=1 
-DHAVE_UNISTD_H=1 -DHAVE_GETPAGESIZE=1 -DHAVE_MMAP=1 -DHAVE_SELECT=1 
-DHAVE_POLL=1 -DHAVE_TM_GMTOFF=1 -DHAVE_INT64T=1 -DHAVE_SOCKLENT=1 -I.. -c ssi.c
 cc -O2 -O3 -march=i686 -O2 -DHAVE__PROGNAME=1 -DHAVE_FCNTL_H=1 -DHAVE_GRP_H=1 
-DHAVE_MEMORY_H=1 -DHAVE_PATHS_H=1 -DHAVE_POLL_H=1 -DHAVE_SYS_POLL_H=1 
-DHAVE_SYS_EVENT_H=1 -DTIME_WITH_SYS_TIME=1 -DHAVE_DIRENT_H=1 -DHAVE_LIBCRYPT=1 
-DHAVE_STRERROR=1 -DHAVE_WAITPID=1 -DHAVE_VSNPRINTF=1 -DHAVE_DAEMON=1 
-DHAVE_SETSID=1 -DHAVE_SETLOGIN=1 -DHAVE_GETADDRINFO=1 -DHAVE_GETNAMEINFO=1 
-DHAVE_GAI_STRERROR=1 -DHAVE_KQUEUE=1 -DHAVE_SIGSET=1 -DHAVE_ATOLL=1 
-DHAVE_UNISTD_H=1 -DHAVE_GETPAGESIZE=1 -DHAVE_MMAP=1 -DHAVE_SELECT=1 
-DHAVE_POLL=1 -DHAVE_TM_GMTOFF=1 -DHAVE_INT64T=1 -DHAVE_SOCKLENT=1 -I.. -c 
match.c
 gcc: match.c: No such file or directory
 gcc: no input files
 *** [../match.o] Error code 1
 1 error
 
 make: stopped in /usr/pkgsrc/www/thttpd/work/thttpd-2.25b/cgi-src
 
 
 Note what happens when actual build starts: make starts building
 from cgi-src subdirectory.  Examination of Makefile.in shows this
 (again, in short):
 
 ALL =          thttpd
 ...
 all:           this subdirs
 this:          $(ALL)
 thttpd: ...
 ...
 subdirs:
        for i in $(SUBDIRS) ; do ( \
         ...
 
 Note empty dependency list for "subdirs" target.  
 But "subdirs" require match.o!
 This is stated in cgi-src/Makefile.in
 
 As I understand, make(1) ordered targets arbitrarily, as
 described in Makefile, but this description is wrong in 
 its incompleteness.
 
 From ident output I see that make has been changed since 3.1,
 thus it may be related to make. Is it bug in make?
 Should I provide more information?