From: David Abrahams (dave_at_[hidden])
Date: 2003-11-18 22:30:26
"Peter Dimov" <pdimov_at_[hidden]> writes:
> Martin Wille wrote:
>> [...] grammar objects
>> would simply use an alternative_thread_specific_ptr instead
>> of all the object_with_id stuff, [...]
>
> The right thing to do here is to just fix thread_specific_ptr (starting with
> its specification, if it precludes the alternative implementation). There is
> no need to call the fixed version alternative_thread_specific_ptr, unless
> the alternative version fails some of the tests where the original
> implementation succeeds.
>
> Incidentally. This case is a good illustration why we should have a formal
> process at Boost that allows contributors to submit papers (and issues) that
> meet certain criteria and guarantees that these papers _will_ be reviewed by
> lists members and the respective library maintainers and that a formal
> resolution _will_ be reached and documented.
I rather agree. Boost.Threads is languishing and its design has
frankly become too important to be allowed to do so.
-- Dave Abrahams Boost Consulting www.boost-consulting.com