Re: internal handling of views

On 2006-08-06, HansF <Fuzzy.Greybeard_at_gmail.com> wrote:
> On Sun, 06 Aug 2006 10:39:12 -0500, Charles M wrote:
>
>
>> Im not sure what you mean here by conceptually true. If its true then updating
>> the table source would indeed have an instantaneous effect on the view
>> (because the data element is the same).
>
> Yes. That is, in fact, what happens.
>

<snip>

>
> A view is nothing more than a STORED SELECT. It is evaluated at run time.
> Since the details are stored, the optimizer may elect to merge the
> definition of the view with other information, should the view be used in
> a compound situation, such as a join ... it is possible that the original
> view may not 'executed' in such a situation.
>
> There is no 'data' attached to the view. It is in the table (and only in
> the table).
>

<snip>

>>
>
> All of this is in the concepts manual, and can easily be proven through
> simple tests.
>

Thanks for the reply Hans. The problem I have with the concept manual is that it states that a view can be THOUGHT OF (emphisis mine) as a stored query or virtual table. Well, when someone says something can be thought of, they usually mean thats not what it truly is, or they would have use IS instead of THOUGHT OF. So, I'm unsure if the concept manual is just talking this tact from there on forward or if they really, really mean thats what it is.

Also in discussing mechanics of a view the manual says that sql involving a view is merged with the views definition (its stored sql). Well, thats all fine and for selecting from a view, but how does it work for updating a view? You can't do an 'update .. select from' if the from doesn't contain any data. It would have to do (possibly complex) logic to figure out where the elements go, not merely merging a select query. On the other hand, the pointer description works just fine for this, and as best I can tell gives all the other characteristics of a view as well. But I've only run across the pointer definition once (read somewhere or told by a DBA - don't recall) and every book I've seen since mentions the stored query definition. So I'm faced with an answer that makes more sense but from a source that may be suspect, or the more prevalent definition.

Anyway, thanks again for the reply.

CMM Received on Mon Aug 07 2006 - 08:17:44 CDT