867
Subj : Re: BC5.02 STL question
To   : borland.public.cpp.borlandcpp
From : Ed Mulroy [TeamB]
Date : Mon May 31 2004 12:22 pm

I think you have a trivial advantage from using an iterator.  However
there is something in your code which makes it slower than it could
be:

> int i;
> for (i = 0;  i < stuff.size();  i++)
> {
>         --- do something with stuff[i] ---
> }

You are calling size() in every iteration of the loop.  Instead do
this:

------------------
int i, n;
for (i = 0, n = stuff.size();  i < n; ++i)
{
        --- do something with stuff[i] ---
}
vector<MyStuff>::iterator i = stuff.begin();
vector<MyStuff>::iterator ender = stuff.end();
while (i != ender)
{
    --- do something with *i ---
    ++i;
}
------------------

These loads stuff.size() or stuff.end() only once.  They also changes
the post-increment to pre-increment.  You are not using the value
prior to incrementing and that is most inefficient.  For an int the
compiler knows about how people do this so optimizes it to ++i but if
i were an iterator or other class item it is inefficient.

If you use the index and don't care about the order (front to back vs
back to front) you can also do this for a (possibly trivial) gain:

------------------
int i;
for (i = stuff.size();  i--; )
{
        --- do something with stuff[i] ---
}
------------------

and if you don't need 'i' later

------------------
for (int i = stuff.size();  i--; )
{
        --- do something with stuff[i] ---
}
------------------

..  Ed

> MarvinAvery wrote in message
> news:40bb312a$1@newsgroups.borland.com...
>
> I have an STL question and since I am using BC5.02, I
> thought I'd ask it here.
>
> I have a vector:
>     vector<MyStuff> stuff;
>
> I wish to traverse the vector. Which is better-- use an
> index:
>
> int i;
> for (i = 0;  i < stuff.size();  i++)
> {
>         --- do something with stuff[i] ---
> }
>
> Or use an iterator:
>
> vector<MyStuff>::iterator i = stuff.begin();
> while (i != stuff.end())
> {
>     --- do something with *i ---
>     i++;
> }
>
> Or does it matter?

.

0