RE: Record seach options - millions of records

I built a program that would parse the description fields and separate each
work and built a file of each word and the relative record number that it
occurred in. The program would then setll by word/reade and chain to the
inventory file by RRN. It was the only way to get all the records when the
descriptions were not formatted correctly. Remember, this was prior to SQL.
The maintenance program for the descriptions would delete all the records
for a specific RRN whenever the description was changed and then reparse the
descriptions.

Tom

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx]
On Behalf Of Paul Raulerson
Sent: Wednesday, May 02, 2007 9:15 PM
To: 'RPG programming on the AS400 / iSeries'
Subject: RE: Record seach options - millions of records

Well- explain then please. Unless you specifically built the file to have
just the records you want in some relative record hash, or put all the
specifc records with 'VENDOR' in them at the front of the file, I don't see
how you did it. Especially on an old slow machine like that. Even snatching
them out of an indexed file would take a few seconds, just based on the DASD
speed.

Again, I am not doubting that you did so, and I routinely do the same type
of thing today, but I have to rig the file to let me do things like that.
And I am doing it on *much* faster hardware and DASD.

-Paul


-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx]
On Behalf Of Tom Huff
Sent: Wednesday, May 02, 2007 10:06 PM
To: 'RPG programming on the AS400 / iSeries'
Subject: RE: Record seach options - millions of records

It didn't have to read the whole million. That's where the application came
in.

-----Original Message-----
From: rpg400-l-bounces+tehuff=attglobal.net@xxxxxxxxxxxx
[mailto:rpg400-l-bounces+tehuff=attglobal.net@xxxxxxxxxxxx] On Behalf Of
Paul Raulerson
Sent: Wednesday, May 02, 2007 8:01 PM
To: 'RPG programming on the AS400 / iSeries'
Subject: RE: Record seach options - millions of records

What machine was this on? I don't remember any AS/400 15 years ago (1992)
that would even READ a million record file in less than a second.
Not saying there wasn't - just curious. :)
-Paul


-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx]
On Behalf Of Tom Huff
Sent: Wednesday, May 02, 2007 9:21 PM
To: 'RPG programming on the AS400 / iSeries'
Subject: RE: Record seach options - millions of records

I wrote an application 15 years ago that would select a few records from a
million record file without a key with sub second response. It is all in how
you program it. This was an inventory file and the sales people had to be
able to select all "valves" out of the million record inventory, while the
customer was placing an order.
Descriptions could vary like
Valve, 1/4 in
1/4 in valve
1/4 in brass valve
Etc.

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx]
On Behalf Of Buck
Sent: Wednesday, May 02, 2007 10:45 AM
To: rpg400-l@xxxxxxxxxxxx
Subject: Re: Record seach options - millions of records

It may be ghosts and goblins on my part from the very distant past to when

a

million records would make a 3x box quiver and blood would be oozing out
from the base of the processor.

We had an online customer history inquiry program that referred to
multi-million-record invoice files and had subsecond response time on
the earliest S/38. Logical files, SETLL, READ.
--buck