From: Matthew Austern (austern_at_[hidden])
Date: 2001-01-16 13:43:45
"John E. Potter" wrote:
> Circular list with no dummy node. Tail points to last node. For
> an empty list end == begin == (0, 0). For a non empty list, end
> == (tail, tail->next) and begin == (0, tail->next). Iterators
> naturally increment to end. There are only n values for current,
> but there are n + 1 values for pred. Iterator== is on pred only.
A good point. That sounds like a useful data structure; it has
all of the advantages of an slist with a pointer to the last element,
and fewer of the disadvantages. (It still has the disadvantage of
increased iterator fragility.)
Performance characteristics will be very different from those of the
slist<> that I currently have. It may be useful to provide both
data structures. At the least, I should do some real measurements.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk