Am Donnerstag, 21. November 2013 um 21:31:58, schrieb Georg Baum <georg.b...@post.rwth-aachen.de> > Kornel Benko wrote: > > > Am Mittwoch, 20. November 2013 um 00:30:54, schrieb Tommaso Cucinotta > > <tomm...@lyx.org> > >> > >> What about boost::interprocess::file_lock ? > >> > >> Note from > >> > >> > http://www.boost.org/doc/libs/1_55_0/doc/html/boost/interprocess/file_lock.html > >> > >> "A file lock can't guarantee synchronization between threads of the same > >> process so just use file locks to synchronize threads from different > >> processes." > > > > Yes, but we need synchronization between different processes. > > > >> Not sure it fits in such case. Otherwise, I'd say the rest of > >> boost::interprocess might fit, e.g.: > >> boost::interprocess::interprocess_mutex > > I'd say it fits perfectly, but I have no experience with it.
Didn't we agree to use as little boost as possible? Memory is fading ...
> > At least lockf() is :POSIX, and it works.
>
> So it is OK for Linux and OS X, but not for Windows. Who is going to
> implement the Windows equivalent? Sorry to play the devils advocate again,
> but I believe these things are important.
ATM, for windows does not change anything.
Until now nobody (on windows) uses cmake tests.
And even if someone does, without parallel calls nothing bad happens.
The only bad thing is, if the test start with fresh user dir.
A workaround would be to call _one_ single export test first.
After that also parallel tests are working.
> Georg
Kornel
signature.asc
Description: This is a digitally signed message part.
