|
Boost : |
From: Kevlin Henney (kevlin_at_[hidden])
Date: 2001-10-18 11:57:31
> From: "Peter Dimov" <pdimov_at_[hidden]>
>From: "Kevlin Henney" <kevlin_at_[hidden]>
>> I should have mentioned why I didn't include this option: historically
>> the original numeric_cast had multiple exception types, but there was a
>> feeling that this was superfluous. I was assuming that the same would
>> also apply to lexical_cast, for consistency. However, I think that now
>> that you have suggested it, it ought to be up for consideration.
>
>Perhaps I need to elaborate. :-)
No, no need to elaborate. What you described was exactly what I was
suggesting, exactly how numeric_cast used exceptions originally, and how
most programmers would expect to use inheritance with exceptions.
My question is that as this fine-grained approach was not favoured
before, should we now reconsider its use. In other words, should we
embrace such an approach more generally, eg for numeric_cast as well?
Kevlin
____________________________________________________________
Kevlin Henney phone: +44 117 942 2990
mailto:kevlin_at_[hidden] mobile: +44 7801 073 508
http://www.curbralan.com fax: +44 870 052 2289
Curbralan: Consultancy + Training + Development + Review
____________________________________________________________
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk