Re: Impossible Database Design?

Nikolai Onken wrote:
> Hey Alvin,
>
> > Agreed, I used CRON as inspiration too.
>
> Did you use a stored procedure to parse the pattern or did you move
> that on the application side?
>

I had to be vendor neutral pl/sql, ms-sql, mysql, hypersonic ... which made my decision easy. I just did it all in the application code.

> the remaining worry I would have is how to ensure integrity...
> What if someone somehow deletes an event through say MySQLAdmin?
> Since the repeating rules are not tied to the actual events (though
> events know of their rule) I would have to make sure that events only
> get added/modified through a clearly defined data storage class or
> so...

Yes.

I just solved the problem with acess via my classes and policies like "no delete access shall be granted". "Hacking is not allowed, otherwise restore from backups".

> Maybe it would be possible to create some sort of checksum after each
> time a pattern changes the entries (what could that consist of? Number
> of rows related to a pattern could be misleading since I theoretically
> could accidentally delete a row and add one)

Sounds too complicated. One strength of the RM is its simplicity, I wouldn't stray from that. And you can't make things fool proof no matter what.

> This especially gets tricky if I want to connect other data to
> events....

Yep, sounds too tricky to me. Avoid complexity and just employ a policy.

> Its quite interesting that there's no commercial main line temporal

I guess it comes down to supply, demand and commercial buzz.

Back in the 90s it was a hot research topic, well at least warm, but from what I can tell when the web and XML came, they plundered most of the temporal reseach.

Spatial databases where also in a similar boat but with lots of geo-nav devices, satellites, cell phones, geographic and demographic applications they've managed to see the light of commerce.

> database existing yet but I'm curious to see what kind of concepts
> 'Temporal Data and the Relational Model' will reveal.

It's too involved to get into here, I tried but it gets too long too quickly.

But you've already discovered some of the problems and you you could see the book's contents at Amazon and you may be able to find some of Lorentzos or Darwen's papers. This material is also covered in some grade courses as well but I'm finding it hard to find any info now, it's lilke a ghost town.

> Are there any
> projects being pushed by bigger companies?

Oracle have extensions for spatial database, web, data-warehousing, XML, object ... but nothing purely in the temporal realm AFAIK.

The temporal db topic intersects with spatial dbs and data-warehousing so you may find stuff burried in there?

> Regards,
>
> Nikolai

Even though it isn't a perfect fit I think you should be able to make good use of a rdbms and arrive at a reasonable solution.

Cheers. Received on Thu May 18 2006 - 07:52:33 CEST