[prev in list] [next in list] [prev in thread] [next in thread] 

List:       boost
Subject:    Re: [boost] Signals Thread-Safety
From:       Doug Gregor <dgregor () cs ! indiana ! edu>
Date:       2005-12-29 22:46:31
Message-ID: 24050de3cb6c94341659925258c6591c () cs ! indiana ! edu
[Download RAW message or body]


On Dec 29, 2005, at 3:25 PM, David A. Greene wrote:
> Is there a desire to make it thread-safe at some point, or is
> it not seen as worthwhile?

Users have been asking for thread safety for years. It is *definitely* 
worth while, and has kept people from using Signals far too often.

>  What would be involved in making
> it thread-safe?  Additional runtime overhead?  Additional
> library code?  Both?  Something else?

There would be some run-time overhead, although it could be factored 
out into a threading policy if it was problematic. Dealing with 
deletion of trackable objects and disconnection of slots in a 
multi-threaded environment becomes rather tricky, both to implement and 
to specify. That, compounded with that lack of time I keep whining 
about, has conspired to keep Signals single-threaded even though we 
should have implemented multi-threaded support a long time ago.

	Doug


[prev in list] [next in list] [prev in thread] [next in thread]