I tried to merge www-client/opera-9.50_beta2 on gentoo freebsd today. It did not work because the ebuild got the workdir wrong. (1) After that the pluginpath patch did not apply. (2) Finally opera did not run because of missing libz.so.2. (3) Then i tried the most recent ebuild 9.50_beta2_p2034 and there i also had problem (2) after fixing that i had problem (3) again and decided to file this bug. I merged with "-qt-static". The opera binary for 9.50_beta2_p2034 misses the following 3 libs: libz.so.2 => not found (0x0) libstdc++.so.4 => not found (0x0) libpthread.so.1 => not found (0x0) Did not check beta2 but there its at least libz.so.2 missing. Reproducible: Always I am going to attach patches for problems (1) and (2). beta2_p* only seem to have problem (2) but i did not check all of them. I could get opera 9.50_beta2_p2034 running after creating symlinks to the missing libs. In fact all of them are present on my system but the names are wrong. # ln -s /usr/lib/libthr.so /usr/lib/libpthread.so.1 # ln -s /usr/lib/libstdc++-v3/libstdc++.so.5.0.7 /usr/lib/libstdc++-v3/libstdc++.so.4 # ln -s /lib/libz.so.1.2.3 /lib/libz.so.2 but i guess these symlinks should be provided by the libs and not by opera.
Created attachment 156393 [details, diff] fix for problem (1) freebsd path ends on tar.bz2 not on tar.gz that is why S is not correct on current ebuild
Created attachment 156395 [details, diff] files/opera-9.50-pluginpath-fbsd.patch new version of files/opera-9.50-pluginpath-fbsd.patch, the old one does not apply
I found a nice workaround for the third problem. Opera can provide symlinks that are only used by opera itself. # ln -s /usr/lib/libstdc++-v3/libstdc++.so.5.0.7 /opt/opera/share/opera/bin/libstdc++.so.4 # ln -s /usr/lib/libthr.so /opt/opera/share/opera/bin/libpthread.so.1 # ln -s /lib/libz.so.1.2.3 /opt/opera/share/opera/bin/libz.so.2 These 3 symlinks did solve problem (3) for me. And with dosym this should be easy to include into opera ebuilds. But i guess the names of the files these links point to should not be hardcoded so i wont write a patch for that.
Er, maybe if you had assigned the bug to the maintainer someone would have responded. Is this bug report still valid?
