From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-04-12 08:50:06
Several posting in this thread have mentioned the standard's vector and
similar at() functions.
It really isn't a good idea to use at() as an example of
anything. Speaking as the person who wrote the at() proposal, I can tell
you that many of the considerations that went into at() had to do with
committee dynamics, timing, or technical issues that don't apply to
anything we will ever talk about at Boost.
For example, indexing was seen as a special case by some who claimed that
indexing errors represented a large percentage (over half, IIRC) of all
errors made by their (student?) users. To try to deal with this concern as
part of an overall error handling strategy would have further delayed the
In any case, the situation we are faced with is so different I don't think
looking back at at() is at all worthwhile. It just confuses the very real
issues being discussed in this thread.
If you are really interested in why at(), or std::string, or similar are
the way they are, ask me privately over a beer sometime:-)
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk