Joe Lewis Wilkins wrote:

> Alain Pastor wrote:
>
> > <big snip>
> > PRINT test$;" Joe's backward method:";FN TICKCOUNT - myTick&;" ticks"
>
> You could have been a little "kinder" and referred to it as "Joe's REVERSE
> method:"   *-}

Hi Joe,

My apologize if I made a boo boo. Is it unpolite to say backward in that case?

>
>
> <more snip?
>
> > Here are the results on my G4/400:
> >
> >                            PPC       68K
> > Old way forward search: 59 ticks   109 ticks
> > Old way backward search: 6 ticks    11 ticks
> > New way backward search: 2 ticks     2 ticks
>
> Very interesting, though it gets a little exaggerated by the extreme length of
> the test str, so the real comparison on a PPC really only shows a difference of
> less than a tenth of a second.  In most cases negligible.

I admit that it could be negligible in real situations, but I like to know that it
is a bit faster.
I tested the same demo prog with a$ = "This is an example" (notice the lack of the
suffix)
the results give:
for both old methods : 16 ticks (PPC) and 28 ticks (68K)
for the new method: 2 ticks (PPC and 68K)
If you increase the size of a$ the amount of time taken seems to increase more
rapidly with the old methods than with the new one, if I don't say something silly
here.
with a$ = "I longer to see this very boring and above all very long sentence reach
its end", I got:
both old ways: 62 ticks (PPC) and 117 ticks (68K)
new way: 2 ticks (PPC) and 3 ticks(68K)

> This NEW syntax, looks
> an awful lot lot C.  I hope we're not going "overboard" in that direction, which
> can easily be done when your boat is only 2.5 meters long. (ted <grin>).

Yes, it is C compatible syntax. I don't like C, but I love the new syntax in FB. Go
figure! Well, at least, it seems to generate more efficient code.

>
>
> Did you make any comparisons using the INSTR function that was proposed by some
> of the other listers? Though I suppose the possibility of multiple "dots" pretty
> much eliminates that approach?

No, I didn't do that.

Cheers

Alain