browse lookup - spped up selection

[GCC Consulting]

> -----Original Message-----
> From: filepro-list-bounces at lists.celestial.com 
> [mailto:filepro-list-bounces at lists.celestial.com] On Behalf 
> Of Dennis Malen
> Sent: Thursday, April 13, 2006 5:56 PM
> To: filepro-list at lists.celestial.com
> Subject: browse lookup - spped up selection
> 
> This probably was discussed before.
> 
> When a browse lookup is used the first part of the setup is 
> to define how you are looking the information in the target 
> file. At that point there is an option for the use of only 
> one index selection.
> 
> The browse lookup utility then allows you to create 
> processing in conjunction with your lookup that has many 
> different criteria. If the processing does not match one can 
> use a drop command etc.
> 
> The problem in that approach is that the lookup takes a very 
> long period of time as every record that meets that criteria 
> is chosen in the index. As every record gets a match there 
> may be 5 other criteria that has to be reviewed in the 
> processing table to determine whether it should be shown on 
> the browse lookup screen or deleted (dropped).
> 
> Using a selection format is obviously a lot quicker.
> 
> Is there any way to speed up the process using a browse lookup.
Dennis,

I use this technique quite often.  Since you haven't posted your processing,
I'll ask a couple of questions.

On what key(s) is your index built?

Are you using and drop all before or drop all after?

Are you using drop all

I read Ken's response and what he is indicating is the sequence of your
drops.  This is very important.

You want to have the most critical drop first.  One that will just drop the
record without further testing.

I have found, for instance, that where I am doing a browse on a particular
person's employee #, using:
If: emp_num lt 19
Then: drop all before;END
If: emp_num gt 19
Then: drop all after;end

After some testing I found that the above is faster then

if:emp_num ne 19
Then: drop all;end

I have timed this and can't tell you why, even with the employee # being the
lookup value, but there is a significant difference.

In my processing, the additional drop criteria have to do with a status
code, age and date last called.

This program is being used through terminal server and the performance is
very good.

If you would like a copy of this, let me know off list.

Richard