>Hi to all who are interested in this topic.
>
>We have had an interesting discussion, including some research, under the topic 'VFP versus C++'. This new topic is kind of a continuation, but it has a slightly different focus. Al Doman postulates the hypothesis that the newer functions are faster than the older functions. See also his post: Re: VFP versus C++ Thread #842594 Message #843390.
>
>Several thoughts come to my mind:
>- Is 32-bit versus 16-bit indeed a factor?
>- Functions are unique, that is, no two functions do exactly the same thing. So, how can they be
> compared with respect to speed?
>- What is the hierarchy of the functions with regard to speed?
>- Can we say that the newest functions compete well with the C++ equivalents?

Peter,

I, for one, do not believe that older functions are slower. 16/32 bit is not an issue since they are recompiled each time we get a new release of foxpro.

The problem you had was the repeated use of at() as in

for i = 1 to n
    p = at('Something', SomeString, i)
endfor

Each time At() is called, it has to rescan the string skipping all previous (i-1) occurrences

So, if the max is n, you will have scanned for 'Something' (n * (n+1) / 2) times.
If n is 10000 : roughly: 10000^2/2 = 50,000,000

Hence the difference in speed. The RegExp and alines() only match an occurrence once

my 2 EUR cents