Re: Another view on analysis and ER
Date: Sat, 08 Dec 2007 03:13:36 +0100
Message-ID: <4759fc87$0$232$e4fe514c_at_news.xs4all.nl>
paul c schreef:
> mAsterdam wrote:
>> paul c schreef: >>> mAsterdam wrote: >>>> paul c schreef: >>>>> <Nomme, Ans> >>>>> <Name, Years> >>>>> >>>>> are different headings. >>>> >>>> Yes. However, having multiple headings in one relation >>>> is not part of RM AFAIK. ... >>> >>> Who said anything about multiple headings in one relation? >> >> I did. It is the way I labelled: >> >> >>> <Nomme, Ans> >> >>> <Name, Years> >> >> ... appearantly not something you intended. >> Why not - or, better: what are they?
>
> Oh, it was you, was it? Whew, that was a close one. Maybe I misled
> with the tilted carets or whatever they're called and should have used
> braces, also by abbreviating them without type names, which seems common
> whenever the purpose isn't affected.
Explicit types do not reduce the number of headers.
> Anyway, I suppose there might not be anything theoretically wrong with
> an rdbms that allowed multi-lingual headings, so that the Frenchman
> could pretend the db was using his lingo and the Englishman his,
> although they might get up to more hijinks whenever RENAME came into
> play than those two nationalities ever did in the last 900 years.
>
> If you insist on such a thing, I hope you'll call them some kind of
> alias as I think there is already more than enough multi-lingual false
> correctness in the world.
Please do not attribute your invention to me just because I labelled it.
This seems like a good place to completely re-quote your claim.
> No need, the RM already is language-neutral, eg.,
>
> <John,33>
> <Johan,33>
> <Jaan,33>
>
> are different, not the same.
>
> Same goes for headings:
>
> <Nomme, Ans>
> <Name, Years>
>
> are different headings.
I read your example as having multiple headings to one relation - which is outside the RM as I know it. You prefer to call them aliases? If so: aliases of what?
> As the Mott's Clamato man said, why stop there? Might as well have
> multi-lingual aliases for relation and relvar names too. What the heck,
> do similar for values in tuples.
Making the values in tuples language-neutral makes sense if the people sharing the data do not share a language.
Exposure: Where I live most people have to learn at least two foreign languages (English and one of French and German).
> For a while, the effect might be
> drastically deleterious for update performance, but eventually an
> optimization theory might appear. So typical of the IT world to
> optimize the tool rather than the problem. The DB world being so
> over-endowed with clarity, I guess adding a good dose of obscurity can't
> hurt it either!
>
> (Just my not too blunt way of doing my bit and helping us all yet again
> that explaining analysis and design is harder than doing it.)
Received on Sat Dec 08 2007 - 03:13:36 CET