Matchtext script results

[J. Landman Gay]

I was going to post a compilation of all the scripts I received in 
response to my matchtext question, but there were so many that the post 
would likely be too long for the listserve. I very much enjoyed looking 
at all the solutions, and was impressed all over again at how many 
different ways there are to solve a problem in Revolution.

What I did was copy each contribution to a test stack, and alter it 
slightly as necessary to fit into this structure:

on mouseUp -- in a button
   findWords "list house dog"
end mouseUp

on findwords pWords
   put "/Users/jgay/Comm Data/Humor/" into tDir
   set the directory to tDir
   put the files into tFiles
   put the ticks into tStart
   repeat for each line l in tFiles
     put url ("file:" & l) into tText
     -- unique handler contents inserted here
     if tMatch then put l & cr after tList
   end repeat
   put the ticks - tStart
   put tList into fld 1
end findwords

If necessary, I did any text replacements before starting the timer 
(replacing spaces with commas, for example) and also included any setup 
stuff before the timer began (putting a string into a "goodchars" 
variable, for instance.) Only the actual file processing was timed. I 
tested with a folder containing 118 amusing text files of varying 
lengths. There were two files in the folder that contained all three of 
the test words. I measured in ticks because a general idea of speed was 
good enough for my purposes.

Here is a summary of the contributions (in the reverse order they were 
posted) and their timings and results on my G5-Intel Mac:

Mark Smith  -- native syntax: 18 ticks. Found 2 matches
Mark Smith -- filter: 18 ticks. Found 2 matches
Mark Smith -- array: 15 ticks. Found 2 matches
Dick Kriesel -- array: 8 ticks. Found 1 match
John Craig -- regex: 242 ticks. Found 2 matches
Dick Kriesel -- array: 8 ticks. Found 2 matches
Brian Yennie -- arrays: 7 ticks; found 2 matches
Jim Ault -- filter: 5 ticks; found 5 matches (strings, not words)
Ken Ray -- regex: 15 ticks (first run),8 ticks (subsequent runs); found 
2 matches
Jacque Gay -- original Rev script: 4 ticks. Found 2 matches

The last one in the list is the one I thought I had to replace with 
something faster. It is simply this:

   repeat for each line l in tFiles
     put url ("file:" & l) into tText
     repeat for each word w in pWords
       put w is among the words of tText into tMatch
       if tMatch = false then exit repeat
     end repeat
     if tMatch then put l & cr after tList
   end repeat

And that is what surprised me -- that no tinkering with arrays, or 
matchtext, or anything else is faster than the most straightforward 
Revolution syntax. I was thinking this would take a long time, but in 
fact it is the fastest way to do it (that I've seen so far, anyway.) 
We've mentioned this on the list before, but I guess I need to be hit on 
the head with the facts occasionally, just to remind me how good we've 
got it.

Surprise. Rev wins out again.

-- 
Jacqueline Landman Gay         |     jacque at hyperactivesw.com
HyperActive Software           |     http://www.hyperactivesw.com