Re: SQL for presentation

Forget about SQL...it has given up on RM for a long time and should not be used as a reference for abstract thinking about RM capabilities. (It may however b used in a pedagogical purpose).

Looking at your exchanges I believe you are asking the wrong questions about the wrong issues.

First, the possibility that an *as is* particular tagetted purpose implementation (HTML) could constitute fundation for some kind of RM implementation model is illusion. HTML was developped and designed to facilitate exchanges on global network. (XML too but in a miserable way!), it was not designed with no other pretention than that. I believe it has done well on such perspective.

Second, constituing relational schemas is only the structural aspect of relation variables. RM impose also to deal with manipulation and integrity. These aspects are totally skipped in the premice you exposed. It is true, that there is *no* reason that an HTML document could not be represented as a relation variable. It may even constitute a complex data type. But that does not say if current HTML structure would allow to support implementation for being operated. If I follow your logic, some important questions that arise would need to be adressed:

1. what kind of HTML directly bound operators would be defined in searching a specific page? (Not a search on attributes but on the HTML document itself.) ? HOw would these operators be implemented both syntaxically and operationnally ?
2. what rules would operate to update an HTML data type given the heteroclite nature of HTML.
3. I could conceive a domain from which HTML values would be drawn but the main problem is that HTML break by design RM rule of data independence. HTML encapsulates (gosh I hate this word) both data (static text) AND operations (behavior such as *font* *bold* ). It would not be unreasonnable that a RM version of HTML would not be HTML at all at design time because of normalization.

Third, I doubt how a language of declarative nature by design could constitute both a manipulation language and a definition language. Several people believe that there is no choice but to create 2 separate languages is a necessity.

frebe73_at_gmail.com wrote:
> > What is a relational version of HTML to you?
>
> In my reply to Kenneth Downs you can find an example.
>
> /Fredrik
Received on Thu Jul 06 2006 - 14:17:47 CEST