From: Chris Gehlker Date: 2002-03-21T23:01:05+09:00 Subject: Re: Ruby/Tk on OSX On 3/21/02 2:08 AM, "MENON Jean-Francois" wrote: > Chris Gehlker wrote: >>> >> >> I don't think so. It's looking for a file named tcltklib.rb > > > I don't think so. tcltklib is a binary library (.so for unix and .dll > for windows) Right you are. Bob wrote: > Tcl is on OSX...Tk is not...unless you get the 8.4a version but I > don't know if that breaks Ruby/Tk or not. I do have the 8.4a version. Here is what's weird from a fresh build: [dslstat473:~] chrisg% tar xzf ruby-1.6.7.tar.gz [dslstat473:~] chrisg% cd ruby-1.6.7/ [dslstat473:~/ruby-1.6.7] chrisg% ./configure --enable-shared --mandir='/usr/local/share/man' ... creating ./config.status creating Makefile creating ext/extmk.rb [dslstat473:~/ruby-1.6.7] chrisg% make ... compiling tcltklib compiling tk cc -fno-common -g -O2 -fno-common -pipe -no-precomp -I/Users/chrisg/ruby-1.6.7 -I/Users/chrisg/ruby-1.6.7 -I/usr/local/include -c tkutil.c cc -dynamic -bundle -undefined suppress -flat_namespace -L/Users/chrisg/ruby-1.6.7 -o tkutil.bundle tkutil.o -L/usr/local/lib -L. -lruby compiling Win32API compiling digest/md5 cc -fno-common -g -O2 -fno-common -pipe -no-precomp -DHAVE_CONFIG_H -I/Users/chrisg/ruby-1.6.7/ext/digest/md5/.. -I/Users/chrisg/ruby-1.6.7 -I/Users/chrisg/ruby-1.6.7 -I/usr/local/include -DHAVE_SYS_CDEFS_H -DHAVE_INTTYPES_H -DHAVE_UNISTD_ H -c md5.c ... [dslstat473:~/ruby-1.6.7] chrisg% sudo make install ... installing syslog syslog.bundle -> /usr/local/lib/ruby/1.6/powerpc-darwin5.3/syslog.bundle chmod 0555 /usr/local/lib/ruby/1.6/powerpc-darwin5.3/syslog.bundle installing tcltklib installing tk tkutil.bundle -> /usr/local/lib/ruby/1.6/powerpc-darwin5.3/tkutil.bundle chmod 0555 /usr/local/lib/ruby/1.6/powerpc-darwin5.3/tkutil.bundle installing Win32API installing digest/md5 md5.bundle -> /usr/local/lib/ruby/1.6/powerpc-darwin5.3/digest/md5.bundle chmod 0555 /usr/local/lib/ruby/1.6/powerpc-darwin5.3/digest/md5.bundle So you see why Jack asked the question originally. Apparently the build process says: compiling tcltklib And installing tcltklib Whether or not the process succeeds. We are supposed to take the absence of any further messages as evidence that it failed, just as we can assume that Win32API was neither compiled nor installed. So now the question is why does the compile of tcltklib fail. My first though was "because it couldn't find the headers". But apparently the compile of tk succeeds and the headers are in the same place. Oh well. I'll hack on this for awhile and get back to you. In the mean time, thanks for the help. Bob's remark that Tcl is there by default is a good clue. -- Laws are the spider's webs which, if anything small falls into them they ensnare it, but large things break through and escape. -Solon, statesman (c. 638-c558 BCE)