lists at futilism.com
Fri Jun 26 17:37:07 CDT 2009
On 26 Jun 2009, at 23:07, Richard Gaskin wrote:
> Ditto what the others have said about the progress bar. Those
> updates prompt so many layers of OS rendering code that they take
> quite a toll. I went to a mod solution with one of my projects and
> it gave me an order of magnitude speed boost.
> But I have a question about the algorithm's logic: if c is
> initialized to 0 but the data in idatalpha is traversed starting
> with the first character, wouldn't the comparison always be one
> character off?
I think you're right - the first comparison would be char 1 of
idataalpha to char 0 (non-existent) f idatabeta, then 2 to 1 etc.
And for any benchmark junkies out there, "add 1 to c" is nearly twice
as fast as "put c + 1 into c" on my machine, though it would only
make a small difference even with hundreds of thousands of iterations
(maybe in this case with large images).
More information about the use-livecode