[tmql-wg] Proposed new requirement: Database implementation

Kal Ahmed kal@techquila.com
Thu, 10 Apr 2003 21:25:02 +0100

To be honest then I think that this is a non-requirement. Actually, its 
worse than that, its and over-requirement - we shouldn't let the TMQL 
work be trapped into supporting a relational model of the SAM any more 
than we should let it be trapped into supporting a hierarchical or 
object-oriented implementation of the SAM. That is probably best done by 
letting the TMQL work be driven by the requirements for the TMQL 
language without consideration to implementation against any model other 
than than the SAM.

I would propose removing this requirement completely.



Lars Marius Garshol wrote:

>* Kal Ahmed
>| It seems to me that this is more an implementation requirement than
>| a requirement of the language itself. 
>Yes and no. It's really about what to do when faced with a construct
>in a potential TMQL that is known to map poorly to existing database
>query languages. Should one try to change it to make it fit? Or should
>one just ignore the issue altogether. This requirement was meant to
>say that we'd try to avoid incompatibilities without saying that we'd
>make that a constraint on ourselves.
>| Why choose SQL in particular ?  Why not OQL ? Why not XQuery ? Why
>| not MyQueryLanguage ?
>It says query languages in general, but singles out SQL because it is
>the most used query language at the moment.
>| The TMQL working group should be experienced and adult enough to
>| realise that they need to create something that is implementable - I
>| don't think that we should pile on them requirements that they write
>| our SQL queries for us too ;-)
>That's pretty much what the requirement is trying to say. :-)