Subject: Re: [boost] GSoC 2010: Heaps and Queues
From: Dan Larkin (danielhlarkin_at_[hidden])
Date: 2010-04-06 11:26:08
Andrew Sutton wrote:
> Pointers are used for updating elements after modifying them. Iterators are
> just abstract pointers (more or less), so really they play the same role.
> They aren't really intended to be used for actual iteration. Besides, you
> actually need the iterators for node-based data structures since T* may
> point inside a node.
Ah, that makes much more sense. I haven't looked too deeply into too
many Boost libraries yet, so I'm not aware - is that iterator
abstraction the preferred method over internal pointer manipulation?
Originally I was considering the node pointers to be strictly internal
structure and something that didn't really need interfacing, but I can
see now why it would be useful, especially when extending the Boost
classes to custom solutions.
> If you plan to put code in your proposal keep it succinct and address the
> code in the text. Basically, leave out the comments :)
I'm used to code being written to speak for itself, but that's for when
it's standing alone, not embedded in a proposal. Point taken. Once
again, about to head into class so I can't take care of the revisions
now, but later tonight I'll get revision 2 worked up and ready. (Also I
apologize that I'm a bit more distracted than I usually would be. This
is an exam week for me.)
Boost list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk