In article <tim.5.721094115@tim.src.utah.edu>, tim@tim.src.utah.edu (Tim Ma) writes:
>
> I have a simple script written in 3.5 that randomizes the order
> >>of records in a file. All it does it import that file as a single TEXT
> >>field, tacks on a random number in anothe field, sorts on the random number,
> >>and then outputs the randomized file. To my surprise, it runs almost 50%
> >>slower in 4.0 thanit did in 3.5 (14 minutes vs. 10 minutes). I'm running a
> >>386SX with 4M of RAM. Any ideas or comments?
>
Why are you sorting a data table anyway? If the reason is to print it in
reports in a sorted manner, do that IN THE REPORT using GROUPINGS! If you need
to sort data in a table, you obviously are not using KEYS. Paradox nearly
requires use of keys for optimum performance. Keys are the trigger to speed
and performance in paradox. I find sorting seldom needed in a data table and I have been building Paradox applications for 5 years. Sorting defeats the
purpose of keys and indexes and should not be used if a well designed database.
Use of INDEXES create secondary keys which can increase searches and queries
performances substantially. Do not use sorts, use Paradox functions for earching tables. Sorts can be used for processing ANSWER tables prior to creating graphs, but that is the ONLY reason I would sort a table...
