atk @2.34.1_1 does not build on PPC Tiger, Mac OS X 10.4.11, because of -rpath

[ballapete (Peter "Pete" Dyballa)]

[44/52] Linking target tests/testrole.
FAILED: tests/testrole 
/opt/local/bin/gcc-apple-4.2  -o tests/testrole 'tests/59830eb@@testrole@exe/testrole.c.o' -L/opt/local/lib -Wl,-dead_strip_dylibs -Wl,-undefined,error -Wl,-headerpad_max_install_names -pipe -Os -arch ppc -Wl,-headerpad_max_install_names -arch ppc atk/libatk-1.0.0.dylib -Wl,-framework -Wl,CoreFoundation -L/opt/local/lib -lglib-2.0 -lintl -Wl,-framework -Wl,CoreFoundation -L/opt/local/lib -lgobject-2.0 -lglib-2.0 -lintl -Wl,-headerpad_max_install_names -Wl,-rpath,@loader_path/../atk
ld: -rpath can only be used when targeting Mac OS X 10.5 or later
collect2: ld returned 1 exit status
[45/52] Compiling C object 'tests/59830eb@@testdocument@exe/testdocument.c.o'.
ninja: build stopped: subcommand failed.
Command failed:  cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_lil.fr.rsync.macports.org_release_tarballs_ports_devel_atk/atk/work/build" && /opt/local/bin/ninja 
Exit code: 1
Error: Failed to build atk: command execution failed

[ballapete (Peter "Pete" Dyballa)]

Main.log from PPC Tiger

[kencu (Ken)]

there are (several?) other tickets about rpath failing on Tiger. We really should condense it all into one ticket.

There are a few options:

build system by build system fixes, port by port fixes are pretty tedious.

I have patched my ld64_97 to not error out when it sees rpaths; this actually works pretty well, because as a final step on install many build systems, like meson, use the full path. So not a breaking tweak, usually...

There is no easy way I know of to allow Tiger to actually support @rpaths in a proper fashion.

Several unix shell scripts are out there that wrap ld to change rpaths to best-guess full paths; these are most likely the best bet, I think.

All a bit hacky, but ...

[kencu (Ken)]

fixed by [4a403a40ba1eb1639e5249d515859c0fb4184c4a/macports-ports]