[Effective-cpp] Item 1: Uses and Abuses of vector

[White Wolf]

effective-cpp-bounces at accu.org wrote:
[SNIP]
>> I thought this a great shame at the time and given the lack of any
>> outcry, took it that the standards committee and other opinion
>> formers were
> broadly
>> in agreement with BS.
> 
> I think you may have misunderstood the scope of it, but
> perhaps members of the committee on this list may fill us in, here.

I think somebody did already, but I cannot seem to find the mail.  Anyway -
as someone who appears at the meetings but *not* a member, yet: as far as my
understanding goes, the main plan is to change the core language to support
library writers to make better libraries and then add most of the the
features as library features.  This of course in no way means that a good
core language extension, with community support and industrial experience is
not considered or ruled out, because it does not fit the
make-life-better-for-library bucket.  Actually, there are (for about a week
now) other buckets.  Herb very nicely details them on his blog:

http://pluralsight.com/blogs/hsutter/archive/2004/10/23/2972.aspx

(why isn't it PHP... ;-)

So as you can see, there is only one bucket, which says that it is looking
mainly for library solutions, and that one is a "sub-bucket" of the Library
working group.  And as you can see the text, it is *very* carefully worded
by Herb not to exclude core language changes proposed by that group, only to
set a principle to prefer library changes: "In this area we want to heavily
concentrate on providing simplification via libraries, so we'd like this to
be led by the library working group."  Of course if LWG finds out by studies
that some things require core language changes, nothing prevents them from
proposing that.

And yes, it is time to propose your "pet-fetures" for C++ but (BIG BUT) on a
way that first please use either comp.lang.c++.moderated, or comp.std.c++
newsgroups, or an approriate ACCU sunb-group or the UK panel etc. to "test"
the reactions of people.  It does not mean that if everyone "hates" your
idea you should not propose it.  It rather means that you should be able to
propose it on a more convincing manner, knowing what people dislike or
misunderstand about it.  It is also possible that while asking for a feature
your will be told an approach that you can already use to get what you want
- without a new feature.  If that does all and does all well, you have
learnt something valuable and time is saved for really needed features.

I am telling the above not to discourage anyone, but to help introducing a
filter/accepted community process to pre-test proposals/ideas.  I predict
that WG21 will have to deal with many-many proposals in the near future.  As
some things have got accepted into the working draft already, some others
are on their way to be, and also the process gets more and more publicity I
predict that the number of proposals will rise in an exponential or even
worse rate.  Unless we introduce an accepted, common way of "normalizing"
the ideas before they hit the desk of the WG21, which has basically only 2
weeks per year to discuss face to face, things may go out of hand even with
the subgrouping.  If nothing else happens but separate people will not
create overlapping proposals, that will already save some time, meaning that
more time can be spent on considering the common-proposal (made by more than
one person).

People in the UK have the great advantage of having a working ACCU as well
as having a working UK C++ panel, "just next door".  This means (as I
understand), that you can get expert feedback on your ideas, even face to
face.  To most of you my dream is a given: to have a C++ standardization
meeting in a real English pub. ;-)

Attila