On 2/11/06, Seth Willits <[EMAIL PROTECTED]> wrote:
> On Feb 10, 2006, at 9:34 PM, Seth Willits wrote:
>
> > 5 minutes and 38 seconds in 2006r1.
> >
> > 5 *seconds* in 5.5.5.
>
> Aha! I should have waited just 3 more minutes to send that email.
> It's the Trim function that's slow. If I simply return line instead
> of Trim(line) it's 68 times faster.
I investigated this a bit more...
It depends on the encoding as well. I've run some tests with diffent encodings:
This is the timing with RB 5.5.5 on a Mac:
4ms ('nil' encoding, RB's Trim)
2ms (macintosh, RB's Trim)
5ms (windows-1252, RB's Trim)
4ms (UTF-8, RB's Trim)
1942ms (UTF-16, RB's Trim)
As you can see, UTF-16 was the only one really bad here.
Now the timings with RB 2006r1:
1ms ('nil' encoding, RB's Trim)
469ms (macintosh, RB's Trim)
715ms (windows-1252, RB's Trim)
7ms (UTF-8, RB's Trim)
6ms (UTF-16, RB's Trim)
Apparently, someone has optimized UTF-16 but messed it up for some others.
Now, before somenone says that the slowdown was perhaps necessary
because the former version was not encoding-savvy with "macintosh" and
"windows" encodings:
I've written my own "FastTrim" method that simply looks at the start
and end of a string (encoding-savvy) and here are the results I get
with it:
15ms ('nil' encoding, FastTrim)
29ms (macintosh, FastTrim)
43ms (windows-1252, FastTrim)
56ms (UTF-8, FastTrim)
25ms (UTF-16, FastTrim)
So, while having an initial overhead, it does not get slower with any
of the more complicated encodings.
Here is a test project, including my RB-written FastTrim method that
helps in some of these bad cases pretty well:
<http://www.tempel.org/ftp/pub/REALbasic/FasterTrim.rbp.zip>
--
Thomas Tempelmann - exaggerating over a million times a day!
http://www.tempel.org/rb/ -- The primary source of outdated REALbasic
plugins and examples
Skype: tempel.org AIM: superTempel
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
