Created attachment 279441 [details] build log Merging libreoffice on ~amd64 with USE -java fails because it doesn't detect junit4. As far as I can see, junit4 should not be necessary since I build with -java. It seems as if configure is called twice, once with the correct parameters and once without parameters, thus enabling java.
Created attachment 279443 [details] emerge --info
See also QA notice at head of log: * QA Notice: USE Flag 'templates' not in IUSE for app-office/libreoffice-3.4.1
(In reply to comment #0) > Created attachment 279441 [details] > build log > > Merging libreoffice on ~amd64 with USE -java fails because it doesn't detect > junit4. As far as I can see, junit4 should not be necessary since I build with > -java. It seems as if configure is called twice, once with the correct > parameters and once without parameters, thus enabling java.. Are you sure that it happens only with -java? I get the same with +java.
I have to agree, I tested it a bit, and it occurs whether java is enabled or not. Also installing junit manually didn't help, the error is still the same. In all configurations the build log looked identical. I changed the summary to reflect this. Also there is a hiccup with zlib earlier in the log. I don't know whether this is important.
Could you please resync and try again, might be another side effect of bug #374243 (the env vars weren't set properly)
(In reply to comment #5) > Could you please resync and try again, might be another side effect of bug > #374243 (the env vars weren't set properly) No difference, unfortunately.
(In reply to comment #5) > Could you please resync and try again, might be another side effect of bug > #374243 (the env vars weren't set properly) Unfortunately it still looks exactly the same.
This is fixed in lo-3.4.2.2
(In reply to comment #8) > This is fixed in lo-3.4.2.2 Really? Well, even if the problem has been fixed, the symptom remains in exactly the same form.
(In reply to comment #9) > (In reply to comment #8) > > This is fixed in lo-3.4.2.2 > Really? Well, even if the problem has been fixed, the symptom remains in > exactly the same form. Oops, sorry...I was wrong, it fails due to the erroneous zlib detection, and no longer re-configures.