Dear Listers,

Unfortunately, Wirt has made two comments on this issue which I feel I must
address.

>B-trees also offer the possiblity of changing the manner by which IMAGE
>databases have been used in the past. In the examples I gave above, AMOUNT is
>clearly a real number. Ordinarily, it would have been considered a little
>declasse to put a hashing index on a real number, but I think that changes
>now. An automatic master could be attached to a real number field, such as
>AMOUNT, and with a b-tree attached, you can now ask range questions and get
>back answers almost immediately.

Even though I couldn't find 'declasse' in my dictionary, I want to point
out to everyone that having b-trees on real number fields or integer number
fields may not be a good idea because these values are not 'hashed' by
IMAGE and can cause serious performance problems in the associated master
datasets if the capacity of the master is not defined correctly after
taking the actual data into account.  I remember dealing with a database
years ago where a key value had a Real datatype and the master ended up
with about 3 primaries and thousands of secondaries hanging off of these
primaries because IMAGE used the right-most 31 bits and in most cases the
key values had the same number represented in these 31 bits.  I have also
seen problems with I4, I2 and I datatypes as well as their J counterparts
so please be careful when adding additional masters to your databases in
order to take advantage of b-trees.

>The addition or deletion of records from the detail datasets will
>proceed as they always have in the past. The appropriate search chains for
>that value in that dataset are simply made longer or shorter. The master and
>its attached b-tree are untouched.

Wirt may have been talking about the fact that no new entries are added to
the master dataset when a record is added to an existing chain, but I must
point out that every time a record is added or deleted from a detail
dataset the associated master(s) are always 'touched' to update the
ChainHead counter and possibly the pointers.

There, I feel better now! :-)

Hope this helps,


+---------------+
|               |
|            r  |  Ken                             [log in to unmask]
|          e    |                           http://www.adager.com
|        g      |  Ken Paul                      Tel 208 726-9100
|      a        |  Customer Support              Fax 208 726-2822
|    d          |  Adager Corporation
|  A            |  Sun Valley, Idaho 83353-3000            U.S.A.
|               |
+---------------+