Re: Evolution of Date's "Abstract Machine" (via CM)

Pete, I don't know what you did to attract the degree of bile you did with this post, but I don't envy you. Thank you for posting new topics and keeping the discussion rolling.

mountain man wrote:
> <SNIP>
>
> It appears to me that Date's "abstract machine" is referenced at
> two separate processes within the prefessional environment of
> database systems, the first being at the commissioning of the
> implementation (in the design process perhaps), and the second
> being during the process of change management --- over the
> years that follow.

Assuming that the database schema is accessible via relations (which is desirable), then it would seem that schema should include a change history.

You could then need constraints between temporally designated application fragments (portions stored in system catalog relations - triggers, stored procs) and their corresponding objects. In short, you'd want a piece of code from a point in time to only "apply" to the other DBMS objects extant at that time.

Although the above does require temporal constraints (overlapping time ranges etc.), it requires only relational theory, albeit at a higher level than application data. Sort of like "higher-order relations," though it's not the same as higher order logics (I don't think). I lack the vocabulary to more accurately designate this.

> <SNIP>
>
> Summarising, theory is just as important, and indeed can be
> just as critical, in change management as it is at initial design,
> and effectively, the "abstract machine" is not a static thing
> but is continually evolved through use.
>
> <SNIP>
>
> 1) Is it clear/true that Date's "abstract machine" necessarily
> evolves due to change management considerations over
> time?

Not exactly. The abstract machine, as I understand it (I don't have his source references), is the relational data model - not a specific data model, for a particular application. So while relational doesn't evolve, my database for my application does.

> 2) Does Date reference this issue?

Not that I know of. I have his Temporal book on my shelf, but haven't read it yet - I think that may offer something, indirectly.

> Thanks for any comments,

I'm not sure I'd thank people before the fact for "any" comments.

• Eric