Calling all bench markers!

Bill wrote:

 > By having it be a full functional app, with certain
 > minimum requirements, an honest comparison can be
 > made regarding code size, OS compatibility, etc.,
 > etc., as well as speed.

Certainly sounds reasonable to me.  The speed aspect
comes from the word counting, and that is very close to
standard ordering benchmarks. So long as there are no
very fast ordering functions in ROM, ordering itself rates
as a decent bench mark.

To size up the problem, let's note that, if you
alphabetically order all words in the file after
uppercasing and count up runs of the same word, then
you have done the job. A better treatment would
probably never create runs of the same word, and that
is maybe enough of a deviation from standard ordering
benchmarks to make one rethink the algorithm.  Still,
if you just hook one of Tedd's code modules for
ordering into Ken's framework you quickly get prototype
that might already be be competitive.

To make benchmark aspect palpable, why don't we post
the King James bible and measure the times for
establishing the the bible's word counts!

Incidentally, re my insistance on files of arbitrary
length I hope that is no sweat; one can just peek at
segments of 32K.  Scrolling is a problem but a
solvable one I believe since we have control of
scrollbar.  This is not a full text editor!!

If you stick with files < 32, most solutions will be
fast enough and FB could only win by an insignificant
margin. Only real-life book-sized (library sized?)
tasks will sort the sheep from the goats.

           Cheers

                 Larry S