From: Frank Mori Hess (frank.hess_at_[hidden])
Date: 2008-04-16 10:59:52
-----BEGIN PGP SIGNED MESSAGE-----
On Wednesday 16 April 2008 10:44 am, Anthony Williams wrote:
> Quoting David Ward <dward_at_[hidden]>:
> > I'm using dynamic linking on windows and since upgrading to boost 1.35,
> > I'm seeing a new dependency on the boost_date_time shared library.
> Yes. The thread library now uses boost.date_time for all the timing
> related features including condition waits, sleeps and timed mutex
> You might get away without needing the date-time DLL: it depends on
> your usage.
> > This isn't a big problem of course, but its nice to be able to ship as
> > few dlls as necessary and we've never had to ship boost_date_time
> > before. Is this something that could be addressed in future versions?
> I'll talk to Jeff and see if we can arrange a header-only subset of
> boost.datetime which won't force the DLL to be linked.
The pthreads version of boost::mutex doesn't pull in date_time though. It
seems like it should be possible to make the win32 version behave similarly.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
-----END PGP SIGNATURE-----