|
Boost : |
From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-11-26 13:12:04
At 11:48 AM 11/26/2001, Jeremy Siek wrote:
>> * A replacement for vector<bool> which does meet container
requirements.
>
>I don't see how the unified class would address this issue. We still need
a
>proxy for the reference type, and therefore it still would not meet the
>container requirements, which is part of my reason for not including the
>container functions in dyn_bitset.
I'm assuming that with the resolution of issue 198 it is now possible to
cache the value (of the correct type) in the iterator, and so the reference
type is not a proxy, but rather the correct type.
But there may be other problems, such as performance, or as you suggest
below, the interface may be conceptually wrong.
>> * bitset isn't an STL container.
>
>Right, and I don't think it should be an STL container. It's an entirely
>different beast.
That still leaves dyn_bitset and bitset to be unified:-)
--Beman
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk