Re: [boost] read_write_mutex fundamentally broken in 1.33?

From: John Maddock (john_at_[hidden])
Date: 2005-09-10 04:36:46

> The company I am working for is actually using timed rw mutex on a current
> project (and they seem to be working), so I for one would like them to
> stay in. I suspect timed locks aren't going to deadlock?

Unfortunately the test case that finally broken the camels back involves 
timed locks on a read write mutex, and yes, timed locks can deadlock 
especially if used in combination with non-timed locks (see 
http://article.gmane.org/gmane.comp.lib.boost.devel/130787 for the original 
message and test case).  Of course this all refers to 1.33 I haven't tested 
1.32.

John.