On Thu, Dec 19, 2013 at 04:36:10PM +0200, Martin Storsjö wrote: > On Tue, 17 Dec 2013, Luca Barbato wrote: > >On 17/12/13 21:06, Martin Storsjö wrote: > >>On Fri, 13 Dec 2013, Luca Barbato wrote: > >> > >>>Module: libav > >>>Branch: master > >>>Commit: a5a3b398fd9dce38ca50b20f182b17a256d209f2 > >>> > >>>Author: Luca Barbato <[email protected]> > >>>Committer: Luca Barbato <[email protected]> > >>>Date: Thu Dec 12 17:05:34 2013 +0100 > >>> > >>>configure: Reorder pthreads checks > >>> > >>>Some pthreads symbols might be present in libc (as shown on various *BSD) > >>>but not all of them, leading to false positives. > >>> > >>>Check for the most common compiler flags before the plain symbol check > >>>to avoid known pitfalls. > >>> > >>>--- > >>> > >>>configure | 8 ++++---- > >>>1 file changed, 4 insertions(+), 4 deletions(-) > >> > >>FWIW, this causes clang/OSX to give warnings when linking about the > >>-pthreads parameter being unused. My vote earlier was on checking for 2 > >>(or more if needed) different pthreads functions to sort out the issue, > >>but if the extra warning is the only issue and others don't think it's > >>important I guess I don't mind. > > > >Sigh, I let this patch and the alternative on review for weeks and I > >wanted not to keep the BSD broken for that long... > > > >The alternative probably can be fixed and used instead if Diego isn't > >that against it. > > > >(clang on linux does accept -pthread obviously =_=) > > So, what do others think, would it be acceptable to check for > multiple functions for this case, since blindly testing flags can > end up with odd situations like this where -pthread doesn't fail but > adds warnings to the build.
Could somebody summarize what the exact issues are on all the platforms? The information is scattered over many messages and commits so that it is hard to follow what the actual problems are. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
