Re: Revised proposal for QUERY DTD

[Vivien Malerba <malerba linuxave net>]

On lun, 10 jui 2000, you wrote:
> 
> Hi all.
> 
> In April 2000 Vivien Malerba sent a proposal to use XML to manipulate queries to this list.
> After asking for her agreement, I made some changes and extensions to the DTD. 
> I added support for joins, unions and ordering and changed the names of the comparison
> operators and some child elements to attributes.
> I made this last change because I don't like constructions like <CONST>123</CONST>.
> IMO it's more natural to make it an empty tag with an attribute <CONST value="123"/>.
> 
> Of course the DTD is not finished now. There should be support for creating, dropping
> (changing)? tables, views, sequences, (triggers, stored procedures)?, more support for
> subqueries etc. 
> 
> Vivien asked me to send the revised DTD to this list for discussion.
> 
> Gerhard
> 

I've had a look at your proposal and here are some comments/questions:
* I think you are right to put upper case keywords for elements and lowe case
for attributes, and also for the renaming of the inf, etc
* Why did you remove the 'id' attribute from the QUERY element? Its purpose was
to give a name to the query.
* Why did you remove the 'allfield' element? Do you prefer to mention all the
fields of a rable rather than using the * symbol?
* Why did you remove the 'aggregate' element? I think it is usefull to make the
difference between aggregates and functions.
* I agree with you on the joins, since you seem to know more about them than me.
* I agree to move the table and view names as attributes, and the const value
as an attribute, it is more logical.
* I agree with your sorting scheme.
* I also agree on the use of NMTOKEN instead of CDATA

Tell me what your opinion is on the questions I have had so far.
Regards,

Vivien