Aye, but the question was which form of repeat was more efficient. In fact your
example proves the point which is, it’s the actual code inside the repeat loop
that takes way more time than the form of repeat itself.
What would really be of interest is comparing the form repeat for each word
theWord of line x of tData to your example. So that is what I did:
on mouseUp
put the ticks into line one of theTime
put "1 2 3 4 5 6 7 8 9 10" into tData
repeat with i = 1 to 100000
repeat with j = 1 to 10
get word j of tData
end repeat
end repeat
put the ticks into line 2 of theTime
repeat with i = 1 to 100000
repeat for each word theWord in tData
get theWord
end repeat
end repeat
put the ticks into line 3 of theTime
put ("Example 1 took " & line 2 of theTime - line 1 of theTime) & " ticks"
into line 4 of theTime
put ("Example 2 took " & line 3 of theTime - line 2 of theTime) & " ticks"
into line 5 of theTime
put theTime
end mouseUp
produces:
83540744521
83540744546
83540744558
Example 1 took 25 ticks
Example 2 took 12 ticks
That is fairly substantial. 25/60th of a SECOND to do 100,000 iterations on the
SLOW loop! And the second form really gets the value twice, but still completes
in 8 ticks. The repeat loop with nothing it it took just over 1 tick. That is
damn fast in my book! Faster than I would have guessed, and much faster than I
will ever need.
I suppose one could argue that some math calculations they can produce would
dwarf my example, but then I would counter that just as LC is not a 3D graphics
game development environment, so also it is not designed to produce lightning
fast math apps.
I’ve likened LC to a constructor set approach to application building before.
As in building a house, if someone wanted complete design freedom to make the
house exactly as they desired, they would want to use raw materials (C++, Java
etc) but if they want to throw a functional house together quickly so they can
get on with building other houses, they should probably get as much pre made as
they can and piece it together. That is LC to me, and thank Runrev for it let
me tell you! :-)
Bob
On Feb 11, 2014, at 09:28 , J. Landman Gay <jac...@hyperactivesw.com> wrote:
> On 2/11/14, 9:48 AM, Bob Sneidar wrote:
>> For small tasks though, your users will not even be able to blink
>> before 100,000 simple repeats are executed. I ran a 100,000 count
>> loop with nothing in the repeat loop to do and it took 1 tick. ONE
>> TICK! The repeat loop is NOT what slows things down!
>
> Well, it depends. Try running your repeat loop with even a single line of
> code inside and use the form "repeat with x = 1 to 100000". That's the
> slowest way to do a repeat loop and I think you'll see a difference,
> particularly if the single line of code accesses any part of the current
> line. Every time through the loop, the engine has to count from 1 all over
> again.
>
> For example, try this and compare with your previous results:
>
> put my100000lineData into tData
> repeat with x = 1 to 100000
> get word 1 of line x of tData
> end repeat
>
> If you get word 10 of each line it will slow down more while the engine
> counts not only the line number but also the word number.
>
> Benchmarks would be interesting, if anyone has time to try it.
>
> --
> Jacqueline Landman Gay | jac...@hyperactivesw.com
> HyperActive Software | http://www.hyperactivesw.com
>
> _______________________________________________
> use-livecode mailing list
> use-livecode@lists.runrev.com
> Please visit this url to subscribe, unsubscribe and manage your subscription
> preferences:
> http://lists.runrev.com/mailman/listinfo/use-livecode
_______________________________________________
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode
