Re: Reusing Deleted Records in a File (REUSEDLT)

[CRPence <CRPbottle@xxxxxxxxx>]

rob@xxxxxxxxx wrote:
> Does anyone process by RRN anymore?  I remember Mapics on a S/34
> used to use "chain chase" sequence on their bill of materials and
> use RRN but I'd like to hope they don't anymore.

  Often such implementations differ little from having an 
/identity/ column [SQL] used as a key.  The same potential for 
problems exist both referentially to data within the database and 
when the data is used outside of the database; e.g. the key value is 
used as a label to identify a physical item outside the database.

  So if an application uses the internal key maintained by the 
database as RRN, there is nothing inherently wrong with that 
application.  The application /can/ make it work.  What is really 
/wrong with/ such implementations, is its _assumption_ that no 
changes to its data would be implemented _outside the application_ 
which might /break/ the application.  The RRN is worse, for its 
inability to be used in maintaining referential integrity by that 
internal key.  The only fatal flaw I can figure with RRN is if the 
physical row is responsible for /damage/ [partial damage] to the 
file, such that a row is not accessible; i.e. it would not be 
possible to simply add a new row as replacement, pending its 
recovery by delete, restore, & apply journaled changes.

Regards, Chuck