-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 03 November 2008 13:54 pm, Johan Råde wrote:
What is the rationale for ditching boost::trackable?
It's not thread-safe. I'm not opposed to including an implementation of it for backwards compatibility though. It does seem like a less painful porting path for single-threaded code should be provided.
If I were to switch from signals to signals2 in the projejct I'm currently working on, some trackable declaration would have to be replaced by several track calls. That would add a lot of clutter and I feel the new mechanism is more error prone.
Wrt clutter and error-proneness, one idea (I haven't decided if it's a good one or not yet) would be to add a signals2::trackable which is based on enable_shared_from_this. It could use visit_each like Boost.Signals does, and then use enable_shared_from_this on any signals2::trackable objects it finds to get a shared_ptr for tracking. Or it could throw an exception if it finds a signals2::trackable object not owned by a shared_ptr. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFJD1ih5vihyNWuA4URArq5AJ9xOZhQPq4JpupLUyj3NOBn+Q/MWwCcDegJ OCa1JkjAVWuESicUBGaRDYk= =IPME -----END PGP SIGNATURE-----