|
Hello Martin,
This is also strange. I guess FreeBSD has a different linker from the GNU tool set, because on my Linux system, everything builds perfectly well. Can I trouble also to put this patch in the bugs database? Sorry for the inconvenience. Thanks, Kern On 04/03/2014 06:57 PM, Martin Simmons wrote: On Thu, 3 Apr 2014 15:27:20 +0100, Martin Simmons said:When building 7.0.2 with --disable-libtool on Linux and FreeBSD, linking bls (and some other tools) fails like this:/usr/bin/g++ -L../lib -L../findlib -o bls bls.o block.o block_util.o butil.o device.o dev.o os.o file_dev.o tape_dev.o label.o match_bsr.o vtape_dev.o ansi_label.o ebcdic.o lock.o autochanger.o acquire.o mount.o parse_bsr.o record_read.o record_write.o record_util.o read_records.o reserve.o scan.o stored_conf.o spool.o sd_plugins.o status.o vol_mgr.o wait.o \ -lbacfind -lbaccfg -lbac -lm -lpthread -ldl -ldl ../lib/libbac.a(message.o): In function `generate_daemon_event(JCR*, char const*)': /tmp/bacula-7.0.2/src/lib/message.c:116: multiple definition of `generate_daemon_event(JCR*, char const*)' bls.o:/tmp/bacula-7.0.2/src/stored/bls.c:29: first defined here collect2: ld returned 1 exit status The problem is that generate_daemon_event is defined in lib/message.c and stored/bls.c (plus some other standalone tools). It worked in 5.x because the other definition was in lib/pythonlib.c, which was not linked into bls etc. Does generate_daemon_event have any purpose now that python is no longer used? If not, perhaps it can be removed completely?Please consider the attached patch, which removes the dummy definitions from the files that don't need them (and probably never did). __Martin |
------------------------------------------------------------------------------
_______________________________________________ Bacula-devel mailing list Bacula-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-devel
