Boost logo

Boost :

From: Peter Dimov (pdimov_at_[hidden])
Date: 2002-02-06 10:52:12


From: "Rainer Deyke" <root_at_[hidden]>
> From: "Peter Dimov" <pdimov_at_[hidden]>
>
> > Right, the standard headers have the wrong granularity. Fine-grained
> headers
> > don't have this problem. You know that std::allocator is in
> <std/allocator>,
> > and that std::multimap is in <std/multimap>.
>
> And 'std::endl' is in <std/endl> and 'std::hex' is in <std/hex>.

There is a break even point where the 'intuitive entity to header mapping'
gets overridden by other considerations, true. This particular case is easy:
we put the manipulators in <ostream> and <istream> (free functions taking
class arguments are logically part of the class.) std::for_each is harder;
<std/for_each> and <algorithm> are the only reasonable choices.


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk