Re: A DBA perspective.

[David RR Webber <Gnosis_@compuserve.com>]

Message text written by Michael Rowley
>
David,

I think you missed my main points.

1) Details matter.  I would like to see what the arguments would be to 
the procedures you mention, so that I can see what queries are possible 
with those procedures.

>>>>>>>>>>  OK - I'm actually just thinking <findQualifiers> at this point
            a la UDDI - and the way a method works - you just
            pass in a valuelist of business content you are wanting
            to locate.   That being said - all the mechanics of the
            actual query interface are inside the method blackbox.
            There is no query syntax mechanics as parameters.
            This is also the familiar model for a SQL proc - all
            the query syntax is in the proc and therefore secure.

2) Being able to query against arbitrary tag names (one of your 
locators) seems to contradict everything you say about limiting what can 
be queried and being independent of implementation.  I suspect that I 
misunderstand what you mean with that.

>>>>>>>>>> OK - this may bust the traditional SQL model since SQL
           needs everything in columns and tables.  
           In an XML model or an HTML search model its much 
           more familiar method.  Search the entire information space
           for occurances of... Oracle could do this if you
           stored everything as XMLBlobs.  At this point though 
           I'm already thinking of rolling this up into an API
           method instead - therefore this would go in the
           'optional' list, since only an XML based implementation
           could actually do this -maybe - I'm sure someone will
           now post the obvious rejoiner that with pre-processing
           of the entire XML content into a single search index
           the Oracle database can do this - but that's a stiff
           bar to raise here - so 'optional' is the correct 
           designation here it would seem.
              

I have not yet argued against the approach that you are trying to put 
together.  I think it _may_ be better than the OQL approach (although I 
prefer something based on XPath), but I want to see more details for the 
approach you are proposing.

>>>>>>>>>>> Thanks - OK - with that encouragement - back to the API - WP!

DW.