Re: [Boost-users] [signals] Automatic connection management and MT

From: Frank Mori Hess (frank.hess_at_[hidden])
Date: 2007-02-06 10:11:42

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 05 February 2007 19:42 pm, Peter Dimov wrote:
> The original code also doesn't extend to derived classes if your program
> takes advantage of the MT-ness of the signal. There is a window after
> the connection is made in which another thread can invoke the signal and
> call the slot, accessing the partially constructed object.

I didn't consider that.  If a class contains a subobject which creates its 
own thread and emits signals, the subobject's signals can't be safely 
connected to the containing object in its constructor.  Boost doesn't 
provide any nice little post-constructor framework by chance?

- -- 
Frank
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFFyJqv5vihyNWuA4URAtvuAJ9k0ojelXsHXqIMWJPEnoJywP7GvwCeN3az
edTd8EVontmaBImvVv7yLEY=
=7xCn
-----END PGP SIGNATURE-----