James Kanze
11/21/2008 10:09:00 AM
On Nov 20, 2:42 pm, Rune Allnor <all...@tele.ntnu.no> wrote:
> I have a function that takes an element in a vector as
> argument. The naive interface goes as
> float computeSomething(const std::vector<float>& v, size_t i)
> {
> size_t j = i-1;
> size_t k = i+1;
> if (j<0){/* Handle start boundary condition */}
> if (k>=v.size()){/* Handle terminal boundary condition */}
> }
> I would like to implement this in terms of iterators. But it
> seems one needs several iterators to get this done:
Yep. That's part of the design of iterators---never use one
parameter when two can do. It makes a lot of things very
painful.
> typedef std::vector<float>::const_iterator pos;
> float computeSomething(pos i, pos b, pos e)
> // Call function with b=v.begin(), e=v.end()
> {
> pos j = i; --j;
> pos k =i; ++k;
>
> if (i==b){/* Handle start boundary condition */}
> if (k==e){/* Handle terminal boundary condition */}
> }
That's more or less an exact equivalent of your original
version, yes. Now if you make your function a template, it will
work with any sequence which is defined by bidirectional
iterators.
> So my question is: Do there exist generally recognizable
> invalid values for the std::vector<>::iterators? Like
> std::string::npos () ?
No.
> If yes I could make a greatly simplified (as well as
> safer) interface to the function:
> float computeSomething(pos i)
> {
> pos j=i;--j;
> pos k = i;++k;
> if(j==std::vector::npos){}
> if(k==std::vector::npos) {}
> }
In other words, you want the iterator to encapsulate both the
current position and both upper and lower bounds somehow. It's
certainly possible, but I'm not sure what you'd gain by it.
> But I can't find any analogy to std::string::npos defined for
> vectors. Does such a value exist?
No, although you could probably define one arbitrarily.
Something like static_cast< size_t >( -1 ). Like
std::string::npos, it would be an index, however, and not an
iterator.
--
James Kanze (GABI Software) email:james.kanze@gmail.com
Conseils en informatique orientée objet/
Beratung in objektorientierter Datenverarbeitung
9 place Sémard, 78210 St.-Cyr-l'École, France, +33 (0)1 30 23 00 34