[boost] Re: what happened to allocators in boost?

From: David Abrahams (dave_at_[hidden])
Date: 2003-08-27 18:42:10

"E. Gladyshev" <egladysh_at_[hidden]> writes:

>> > BTW:
>> > I'd a bit suprised if the C++ committe
>> > accepts Boost memory management concept 
>> > (or a complete lack of such) as 
>> > an industry standard.
>> 
>> Too late (sort of)!  Have you looked at TC1?
>> [...]
>
> It means that we are screwed again
> for a long time to come.

For some definition of "we", I guess so.  I guess an implementation of
C++ tailored for embedded platforms is free to supply
allocator-paramterized components as an extension, though.

> First, it was an inconsistent and incomplete 
> standad for STL allocators, now it is Boost and
> STL combined and things are getting even worse.

Consider: yours is a problem that nobody on the committee is convinced
they have a good solution for.  Any library design will fail to serve
some specific need that seems related to the library's purpose.  In
your case, "a generic callback design with full control over memory
allocation" is one such need.  Accepting the TR1 proposals benefits a
great many users, and rejecting them because they fail to benefit a
few others is no help to anyone.  It would be even worse to try to
throw a half-baked solution which nobody is convinced is right (like
adding an allocator parameter to everything) at the problem.

On this note, some of Andrei Alexandrescu's policy-based designs seem
to provide all of the paramterization you could require.  Though some
people have complained bitterly that we don't have such components in
Boost, nobody has brought one forward for formal review.  Maybe you
should think about doing that yourself.

> As soon as it is an official standard,
> people won't be motivated to redesign 
> their standard libraries in the near future.
>
> So there is not much point to discuss it now.

No, there are quite a few years before TR1 becomes standard.  If you
can present a comprehensive vision of how your problems should be
addressed it's certainly not too late to have an impact.  I mean this
in the most positive way, though I can't think of a better way to say
it: that would be much more productive than griping that others
haven't solved the problem for you.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com