Magic Mailing List |
|
From: number9 AT plate DOT dimlight.org Date: Thu Apr 15 2004 - 13:41:54 EDT
Following the advice of Michael given below earlier in the list, I was able to get Magic working fine on my PowerBook G4. I wanted to let the list know, as I had some serious problems getting it to work before... A special note here is that it would not work (correctly/fully) under OS 10.2, but after I installed "Panther" everything compiles and works fine. Thanks for all of your help. Robert number9 AT dimlight DOT org On Fri, Mar 26, 2004 at 09:33:27AM -0500, Michael Bajura wrote: > Hey Tim, > > Great to hear from you. I was wondering whether anyone was out there after > I got several bounced messages from full mailboxes. > > > > There was also one other fix in scripts/configure: > > Around line 8200 in the "darwin" section, it adds the flag "-bind_at_load" > to LDDL_FLAGS. This flag isn't recognized either. I am not sure whether > it's needed or what it does. The only change I tried was "-noprebind" > instead (the most similar sounding option on the man page). After that > everything linked and I was able to get magic started. > > Changed from: > LDDL_FLAGS="-dynamiclib -flat_namespace -undefined suppress > -bind_at_load" > Changed to: > LDDL_FLAGS="-dynamiclib -flat_namespace -undefined suppress > -noprebind" > > > I am running MAX OS X 10.3.3. > > Thanks, > > MIke > > On 3/26/04 9:03, "R. Timothy Edwards" <tim AT stravinsky DOT jhuapl.edu> wrote: > > > Dear Mike, > > > > Thanks for the feedback. Here's the response: > > > >> 1) I seem to be getting errors because there are extra spaces around the > >> "=" sign on the LDDL_FLAGS line, I also seem to be getting errors because > >> there are no "'s in the LDFLAGS line. > > > > Okay, that's easy to fix. > > > >> 2) Also, later in the script there's an unconditional test for $GCC > >> Which unconditionally adds > >> "-Wl,--version-script=\${MAGICDIR}/magic/symbol.map\" on to the end > >> of LDDL_FLAGS. > >> > >> Apparently the compiler that comes with mac OSX 10.3 is already called "gcc" > >> and does NOT accept these options. Maybe this test should be under the > >> "darwin" test earlier in the script? > > > > Actually, I worked on this problem Wednesday because the same thing happens > > on Solaris. The problem is that the compiler is "gcc" but those particular > > flags are sent to the linker. The linker is called "ld" (in Solaris, at > > least) but GNU's "ld" takes different arguments than the Solaris "ld". I > > assume the same thing happens in OSX. A fix is in the "current revision > > 64" (I updated without changing the revision number because I thought the > > issue was only restricted to a handful of Solaris systems). However, I > > am not sure what options need to be passed to the OS-X linker when that > > linker is *not* GNU ld. Hopefully somebody can clue me in on this. > > > > Regards, > > Tim > > > > > MIke > > -- > Dr. Michael Bajura, Computer Scientist > Information Sciences Institute > University of Southern California, Viterbi School of Engineering > 3811 Fairfax Drive Suite 200 > Arlington, VA 22203 > > Tel 703-812-3719 > Fax 703-812-3712 > Email mbajura AT isi DOT edu > > >
|
|