Date: 2001-07-31 10:40:07
--- In boost_at_y..., "Philip Nash" <philip.nash_at_c...> wrote:
> Bill Kempf said:
> > a null_mutex type can be coded up with out any effort at all
> > For that reason I've not included this type [null_mutex] in
> > important that we get Boost.Threads accepted quickly, and this
> need to keep
> > the library "minimal but complete" for the initial submission.
> I understand the reasoning as presented but I think that is a
shame. I know
> a null_mutex is trivial to write, because I have written plenty :-)
> be nice when we can just use the one that is there (although, by
> all alternate "implementations" should be interchangeable anyway).
It's trivial to write, yes, but it's complicated to get it exactly
right as a general concept instead of a concept for a specific use.
In other words, for inclusion in Boost.Threads proper we have to take
into consideration such issues as compatibility (specifically lack
there of) with boost::condition, etc., while usage in a library such
as the sockets library under discussion can ignore such issues since
they're an implementation detail. I don't want the minute details of
such a simple concept to cause any possible delay. So I understand
why the decision to leave this (and several other concepts) out is
dissapointing to many, but that's just going to be the way it is. :(
> Anyway, what is the status of this submission now? I keep reading
that it is
> "nearly ready" - is that still the case or have I missed something?
It's very close to being ready. Beman and I are currently working
hard on the documentation and cleaning up some issues revealed by
this work. I'm also considering a rework to the tss type that will,
unfortunately, take a week or two to get right. All that said, I
expect to be ready to submit within a few weeks, barring any
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk