Automatic vs. demand indexes

[Nancy Palmquist]

fp at casabellagallery.com wrote:
> Nancy posted:
> 
> [...]
> 
>>However, I had cause to try the exact same thing with a Demand index. 
>>Built on the same two fields: state & city
>>Built with the same two lengths: 2 & 8
>>
> 
> [...]
> 
> I always thought demand indexes were really meant to be
> used for reports not browse.  Truth is that I have never
> used them.
> 
> I do have a couple of questions for you:
> 
> Q. Don't demand indexes need to be rebuilt every time you
>    invoke them?  If not rebuilt, don't they have only data
>    as per last built?

You are certainly right.  But in the case where the file does not change 
very rapidly, it works great.

> 
> Q. If they do need to be rebuilt, do they actually rebuilt
>    faster than using 'drop' within your code?

YES, YES, YES - I can rebuild the list once in a while when the subset 
changes.  Rebuild time(1 min tops).  To use drop to look through 15,000 
records just to find 100 records, is very slow (3 minutes before it 
would display the browse) and there was no way to index the file, 
dynamically, to improve that speed.
> 
> It never occurred to me you could use demand indexes this way
> mainly because of my stated understanding (them having to be
> rebuilt).

I have used them to generate an index of selected items for a report. 
Then the exact same items can be edited, selected for subsequent 
reports, posted, etc. by just using the index created.

I have used them to subset customers, inventory or price lists.  They 
cause no overhead in those files since they are not dynamic and never 
opened unless you reference them.

Hope that gives you some ideas.

Nancy

> 


-- 
Nancy Palmquist 		MOS & filePro Training Available
Virtual Software Systems	Web Based Training and Consulting	
PHONE: (412) 835-9417		   Web site:  http://www.vss3.com