#10944 closed defect (fixed)
BUG: net/samba3 builds bad libsmbclient.dylib
Reported by: | ecronin (Eric Cronin) | Owned by: | mww@… |
---|---|---|---|
Priority: | Low | Milestone: | |
Component: | ports | Version: | 1.3.2 |
Keywords: | Cc: | ||
Port: |
Description
The samba3 configure script sets up the shared lib linking stage to use -bundle instead of -dynamiclib, resulting in a libsmbclient.dylib which, at least in 10.4 ppc, the linker dislikes when compiling programs depending on samba3.
The patch attached uses -dynamiclib and the required -fno-common, producing a valid library. Unfortunately, the library thinks its bin/libsmbclient.dylib in otool -L of programs linked against it, which causes a run-time failure when it can't be located. This is where my knowledge of darwin/mach-o linking reaches its end, so I leave this as a bug and not a patch.
MPlayer 1.0rc1 with the --enable-smb flag to configure (working on an updated Portfile for it right now) is what uncovered this if you need a regression test.
Attachments (2)
Change History (7)
Changed 18 years ago by ecronin (Eric Cronin)
Attachment: | Portfile.diff added |
---|
comment:1 Changed 18 years ago by ecronin (Eric Cronin)
comment:2 Changed 18 years ago by pipping@…
Milestone: | → Port Bugs |
---|
comment:3 Changed 17 years ago by ecronin (Eric Cronin)
Portfileb.diff is a second patch (requires the first) which fixes the unresolved references to e.g. 'bin/libsmbclient.dylib' for libraries in /opt/local/lib/samba3, but libraries in /opt/local/lib/samba3/somedir are still broken (I had to manually specify the install directory at link time, and the samba Makefile does not provide any distinction based on where libs end up).
Changed 17 years ago by ecronin (Eric Cronin)
Attachment: | Portfile_b.diff added |
---|
comment:4 Changed 17 years ago by markd@…
Resolution: | → fixed |
---|---|
Status: | new → closed |
Fixed (and updated) in r25758.
Also, this is pretty clearly a bug with samba and not mports/dports, but if we could give them a complete patch instead of just a bugreport I think it'll get fixed upstream faster...