Hi all, it has been a while since we first encountered the `-lpthread issue', as summarized[1] by Thomas Schwinge. 1. http://nic-nac-project.de/~schwinge/tmp/libgtkspell Some workarounds have been or are about to be used to get packages compiled, e.g. the trick I suggested in the mkspecs/hurd-g++/qmake.conf file, in a qt3 patch, on alioth (sorry, no direct link, I'm in the train), which consists in defining some constants strangely. That's something like defining: LIB_FLAGS = -lpthread THREAD_LIB_FLAGS = instead of the opposite... I used that because looking at the source of qmake (IIRC), the first variable is appended to the compilation/linking flags before the second, which makes `-lpthread' be the first `-l' flag, and makes the build succeed, as well as the execution of the program (qt3 build process is heavily based on qmake and is successful). That's definitely both ugly and _not_ robust. And since I don't want to see that problem being forgotten and then discovered again, after some painful and fruitless debugging hours, I'm typing this (not so) tiny reminder mail. I guess that opening a bug on alioth would be The Right Thing© but I can't do that at the moment, since I'm going to travel alot during the weekend. Sorry for being so repetitive on that topic, but IMHO that matter should be addressed as soon as possible, since threads are used very often, as well as libraries based on threads, like qt3 and qt4 ones. Thanks for your attention and sorry for the inconvenience. Best regards, -- Cyril Brulebois PS: I wondered whether I should Cc: that mail to bug-hurd, please forward it if you think so. Being unsure, I don't.
Attachment:
pgpYDry00v6y9.pgp
Description: PGP signature