Re: Queue

["Marin David Condic" <dont.bother.mcondic.auntie.spam@[acm.org>]

Well, I guess the problem I see is that we could easily keep going back and
forth on what constitutes the best possible solution indefinitely. We'd be
far better off getting something as an answer - even if it is imperfect.

Maybe its hopeless? Is it a little too much like asking "What's the best
kind of car to drive?" or "What's the best shotgun to have?" - Too many
potential uses and personal preferences to come up with a "best" answer and
nobody wants to settle for a sub-optimal answer. Maybe we'd end up with
Lists and Maps if the next standard just has them show up in a sneak attack,
but it seems that this process is not likely to get a consensus strong
enough to get a result.

I think that Ted proposed a package that would make a good start. But that
rather quickly shot off in a few dozen directions with objections on all
sorts of fronts - and it was only one little lonely list package.  I could
still imagine taking it pretty much "as is" and adding to it a map package
that looked similar and declaring that "Version 1.0". Any chance that
proposal would gain any support? Anybody else have a proposed answer?

I think if we don't commit to some direction pretty soon, the stagnation and
deadlock will result in a lack of interest and this will wither on the vine.
Too bad. Its a good idea and a noble goal.

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com
Web:      http://www.mcondic.com/


"Matthew Heaney" <mheaney@on2.com> wrote in message
news:u07tltbt28eq1d@corp.supernews.com...
>
> No.  The problem with the Booch Components is that they use inheritance,
> which is completely unnecessary unless you require dynamic polymorphic
> behavior.
>
> The C++ STL doesn't use inheritance.  The use of iterators and algorithms
> abstracts away the details of specific containers, and thus treats all
> collections as a "sequence."  This is a form of static polymorphism.
>
> If you do need a dynamically polymorphic container (a stack, say), then
you
> can implement that yourself using an Adapter pattern.
>
> The standard library should provide only the most basic primitives, from
> which higher-level abstractions can be constructed.
>
>
>
>
>