Re: some ideas about db rheory

vldm10 wrote:
> Recently I found this article on the internet
> http://www.cs.mu.oz.au/~rui/publication/vldb08_TransactionTimeIndexing.pdf
> and have decided to write a reply on the following Reinier’s message
> from a union is always a join! :
> ...

I doubt if I would get the drift of the rest of your message no matter how hard I tried, so I snipped it, but would like to ask if you would please give a link to Reiner's post about 'union is always a join'. I remember sending a series of messages with that subject. It was a provocation that is slightly reminiscent of the rest of your post, which has to do with 'update history' as far as I can tell. Whereas my motive was to discount history, ie., if a relation's extension can be materialized, then I was hoping that people might see the way clear for an implementation to choose an arbitrary history, but one that would be consistent for all extensions in some sense. This seemed an easy way to get around the so-called view update problem, but I gave up on it as the only person, as I recall, who gave the slightest hint that he got the idea was Bob B and he didn't care to elaborate (which I don't mean as a complaint, just an observation). Received on Wed Jul 08 2009 - 05:30:00 CEST