Re: cdt glossary - TABLE

From: mAsterdam <mAsterdam_at_vrijdag.org>
Date: Thu, 14 Jul 2005 02:15:27 +0200
Message-ID: <42d5ae9e$0$32141$e4fe514c@news.xs4all.nl>

Paul wrote:
> dawn wrote:
>

>>Do you call the result of a SQL/relational query a "table" or just
>>stick with "set"?  I would guess that people would intuitively consider
>>a result set to be a table.  We can define it either way and usually
>>whether it is ordered or not is not relevant, but when we have a need
>>for precision, it would be good to know which is most commonly
>>accepted.

>
> I tend to use the words "table" and "relation" as synonyms when talking
> about databases, mainly for pedagogical reasons; maybe I shouldn't
> though. I just think people new to databases get confused by the term
> "relation" and think it's something to do with the relationships between
> tables - I think I certainly did. If I want to talk about SQL tables
> which may have duplicate rows I'd use the term "SQL tables".

Basically I agree. On a similar note: When discussing ER I tend to avoid 'relationship' and use 'association' (to avoid confusion with 'relation') - at the end I rectify.

However, it's been a while since I adressed a naive audience (i.e. not having ideas, skewed by either Oracle, DB2 or OO - e.g. Hibernate, google for it) - so maybe I should say I used to agree - and would agree if I'ld get the chance again.

>>Do models other than the relational model use the term "table" and, if
>>so, do these models define "table" to be the same as relational
>>theorists define it?  I think of a table as being able to be modeled
>>with a mathematical matrix, but I might not be considered a good
>>relational disciple ;-)

>
> Loads of words are overloaded with meaning so I think we're onto a
> losing battle trying to have definitions that are consistent across
> disciplines!

A valuable database manifests a language or at least a dialect in an organisation. You simply cannot afford to avoid multidisciplinarity. Consistency across may be out of reach, but mutual respect is not.

> A table in relational database-speak could be totally different to a
> table in math-speak or HTML-speak or even SQL-speak.
>
> In most non-RDBMS contexts a table would be considered to have order of
> both columns and rows, I'd say, just like a mathematical matrix. But all
> this is just convention really.

Don't underestimate convention. Received on Wed Jul 13 2005 - 19:15:27 CDT