From: Beman Dawes (beman_at_[hidden])
Date: 2000-11-07 16:49:06
Kevlin,
Regardless of the final outcome, please supply rationale in the docs for
the resolution of these issues. Otherwise they will be revisited
time-and-time-again.
--Beman
>In message <B62D9E2D.14708%darin_at_[hidden]>, Darin Adler
><darin_at_[hidden]> writes
>>on 11/7/00 10:42 AM, Mark Rodgers at mark.rodgers_at_[hidden] wrote:
>>
>>> I have to say that I feel pretty strongly that any should lose
>>> this conversion. It allows comparisons of anys and IMHO that is
>>> a very Bad Thing.
>>
>>I strongly agree.
>>
>>Note that the issue is not merely that it allows comparisons of anys,
but
>>rather than it allows them with bad semantics.
>
>I'm sensing more consensus than we had last time round on this issue, as
>well as more problems with the status quo. Historically I think that's
>the last vestige of smart-pointer-like behaviour in any. I think
>majority and safety should prevail on this one, so the UDC is now
>deprecated (and will not be in the final release) in favour of an
>explicit empty query.
>
>Historically this also turns back the clock a little, as empty was what
>was originally supported, but also raises another question over
>something I removed a while ago: should a clear function be added to
>support the more explicit approach? Ie a.clear() as opposed to a =
>any().
>
>Kevlin