[Tfug] Inheritance in RDBMS's (PostgreSQL, Oracle, et al.)

[Bexley Hall]

Hi Terence,

[snip]

>> What's the downside of using inheritance in designing
>> data models for RDBMS's?
>
> Extremely difficult to  maintain.

Hmmm... that's a biggie!  :>  But, if the model *doesn't*
change (or, if that change is essentially in the form of
a "product upgrade" which automates this for the user...)

> If the base type changes then the
> data would need to be reloaded.  The simple type I showed
> does not have methods but adding verious member functions
> can be very helpful..

OK.  Using "accessors", and the like?
 
> Need to practice good discipline in coding and naming

Ah, I guess keeping track of the various "players" could
become confusing (what's where).
 
>> Conversely, are there any (LESS OBVIOUS) advantages
>> to this approach?
>
> If the base type is well
> defined addition can be easy.  Doing  versioning can
> become a snap  because base columns are there and
> additional columns are there based on version.  

I'm not sure I understand how this is a consequence of inheritance?
Could you elaborate? 
 
>> Any general issues to avoid or exploit to take better
>> advantage of this capability?
> Create constructor and helper methods to make  access even
> smother.  

Thanks!
--don