-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Wednesday 02 April 2003 2:25 pm, Shahan, Ray wrote:
> Hi all,
>         we have some fairly large data sets in terms of number of entries

Are you using the term "data sets" generically, or specifically referring to 
TurboImage datasets?  If you're talking image, there may be a better way to 
do this...

> (49 million +) and just cruising through to get a count of some data type
> is done quite often.

This implies serial reading of the "dataset" -- obviously, if these are flat 
files or other pseudo-proprietary format, this is unavoidable.  If this is a 
true TI database, have you considered making the field a link to an 
auto-master?  [yes, I know someone is going to have a reason this may not be 
a good thing to do, but as they say, every situation is unique...]  since you 
say this is done "quite often", there may be suitable justification for using 
an auto-master.

An auto-master used in this way would be very similar to a plain SQL "index", 
but the advantage here is that you can use some DBINFO intrinsics (or a 
single DBGET for that matter) to determine the chain length instantly rather 
than read millions of records and simply count them if they qualify [and the 
second thought is that if you are counting records that qualify for a reason, 
you probably at some point will want to see those records -- with an 
auto-master you'll have a linked chain for every qualifying record...]

- -- 
Yet another Blog: http://osnut.homelinux.net
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
Comment: http://osnut.homelinux.net/TomEmerson.asc

iD8DBQE+i7jGV/YHUqq2SwsRAm1mAJ9BCBSUS2U94NQi1+xPp3FCRdIB/wCgiy0r
U6a0F1DqOwLLWsPpFaLEYrM=
=DJiU
-----END PGP SIGNATURE-----

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *