From: David Abrahams (dave_at_[hidden])
Date: 2002-10-16 17:19:55
Aside from editorial changes which I'll post as a patch set:
The term "concept" is overused in this document, where "primitive",
"construct", or "object" would be more appropriate.
The rationale for the Lock object is somewhat roundabout. The second
paragraph begins:
"A lock object is not a synchronization primitive"
But you've never defined "synchronization primitive", so this is not
meaningful. The crux of the issue is buried in the least-emphasized
part of the paragraph (the middle):
"instances of a lock object are only going to be created, at least
in theory, within block scope and won't be shared between
threads"
[what's with '? I doubt that's legal, and the job could be done by
"'" just as well, couldn't it?]
But anyway, I'll attempt a rewording in my patch set.
In the sequence of usage comparisons between thread_ref and
noncopyable thread object, it's unclear in some cases which is
which. A side-by-side table with descriptive headings would be easier.
--
David Abrahams
dave_at_[hidden] * http://www.boost-consulting.com
Building C/C++ Extensions for Python: Dec 9-11, Austin, TX
http://www.enthought.com/training/building_extensions.html