Re: row vs row.column level locking ++ constraints and TAPIs

On Sun, 18 Dec 2005 08:08:41 -0500, Mark C. Stock wrote:

> 2) TAPIs -- few applications (should) allow direct end-user DML for any
> significant business transaction. Any database that does not have proper
> constraints (declarative or triggered) and allows carte-blanche insert
> update delete directly against the tables is subject to logical data
> corruption. Additionally, some sort of business logic layer (tier) somewhere
> between the user and the data must be responsible for any data and
> transaction integrity issues that cannot be enforced by triggers and
> constraints -- preferably implemented in database packages and procedures.
>
> So no new issues in these areas if and when 'CLL' is implemented.

Mark, this has been a purely theoretical thread, since nobody is supporting this level of isolation. I've been seeing on this group for a very long time, longer then either of us would like to admit. You strike me as a really serious and experienced DBA. Somehow, I doubt that you will be the first person to have this thing applied to your production database when and if it becomes available. I will wait and see to make up my mind, just like the good folks in 1938 who relied on radio to provide them with information, when Orson Wells begun his historic "War of the worlds" drama.

-- 
http://www.mgogala.com