|
Boost Users : |
From: Kai Brüning (kai_at_[hidden])
Date: 2006-12-19 11:14:26
Hi,
I am new to this list, although not completely new to using Boost.
When working with mutexes and locks, it is common
to have functions which are part of critical
sections but do not lock themselves. A typical
example for this is the thread safe interface
design pattern, which moves all locking to the
publicly available interface functions of a class
and all implementation to internal functions
which expect to be called with a lock in effect.
I would feel much more confident if I could check
"expect to be called with a lock in effect" in
debug builds. That is, if I could test whether a
mutex is locked for a specific thread (in my
case, "current thread" would suffice). Especially
if some functions need locking and others don't
and the goal is to minimize locking for
performance's sake, pure convention and
documentation without runtime checks seems to me
dangerously error prone.
Is there any way to do this test which I overlooked?
Is my desire sensible? Or are there other/better
ways to make sure necessary locking is done?
Best regards,
Kai Brüning
-- RagTime GmbH Tel: [49] (2103) 9657-0 Neustraße 69 Fax: [49] (2103) 9657-96 D-40721 Hilden mailto:kai_at_[hidden] Deutschland http://www.RagTime.de/
Boost-users list run by williamkempf at hotmail.com, kalb at libertysoft.com, bjorn.karlsson at readsoft.com, gregod at cs.rpi.edu, wekempf at cox.net