Re: Setting up updatable views ?

Cimode wrote:
> ... So here's the taxonomy I came up with so far:
> Identification requirements: all requirements that a computing model
> should meet in terms of how such system establishes discernability
> between tuples to support efficiently view updates.
> ...

My two cents on this is pretty humdrum - as long as the Information Principle appears to have been followed, the physical layer can decide what it thinks is "efficient", eg., if it hashing is thought to be physically efficient the implementation is free to hash.

> Predictability requirements: all requirements that a computing model
> should meet to allow *predictable* updates. In such requirements, I
> usually ask myself question such as: how ought a grouped query ought
> to handle to be updated and how would it give a result that would be
> reasonnable and coherent to understanding.
> ...

I think that if the updating tuples are specified in a grouped form then   conventional relational algebra is predictable (predictable in the sense of Codd's model and algebra as it has evolved), unless a view involving projection, union or join is involved. If the updating tuples aren't specified that way, then there needs to be some interpretation of the target, eg., if there is only one attribute in the grouped target, say:

CombinationsShipped
P#

{1,2}
{2,3}
{3}

(3 tuples, meaning one shipment included parts 1 and 2, another parts 2 and 3 and another only part 3), then to insert the fact that every shipment included a free giveaway part 99 can't be done by inserting a single grouped tuple, "insert {99}" if you'll pardon my syntax, even if CombinationsShipped isn't a view.

> Concurrency requirements: all requirements that a computing model
> should meet to allow comit/rollback principle over view operations.
  > Here I ask my self questions such as : can we apply a 2 phase commit > model to hadle concurrency?

I thought it was the 2-phase Lock protocols that were involved in async-transaction/multi-user concurrency within a single dbms, whereas 2PC applied to multiple dbms's, or did you mean distributed dbms's? Received on Fri Oct 05 2007 - 19:33:41 CEST