Hi Mike,
For the sake of argument, I probably could make an argument that the defining of a regular expression is procedural as you suggest I could in your previous two emails, but it is not because users may think of them as being processed from left to right, users don?t care how they are processed internally. The point is that the user had to think at all about how to solve the problem and then have to specify the logic in any way. This is specifying the How of procedural and not the What of nonprocedural.
I also do not believe I changed or switched my argument in my previous email by introducing the SQL example: SELECT c.vals FROM ABCview WHERE B.id=4. I converted your XPath example of A/B[@ID=?4?]C into SQL to demonstrate it being more nonprocedural. But the data independence issue you also bring up can be used to discuss different levels of what can be considered nonprocedural. The same SQL query with the SQLView defining a multi-leg structure, lets say (A directly over B and C) a multi-leg structure with B and C being siblings would also be seamlessly processed correctly. But you claim this is only an orthogonal issue. Then I guess multi-leg hierarchical queries are being performed today in XPath 2.0 and XQuery, and they are being hierarchically performed correctly taking into account LCA (Lowest Common Ancestor) hierarchical processing logic. Please direct me to some documentation showing or explaining this (other than my own).
The real issue of my article was not whether XPath navigation is procedural or nonprocedural. The issue was that user specified navigation of any kind is holding back the database processing of XML and that navigationless processing is needed. This also allows interactivity. This allows unskilled XML users to interactively query XML without concerns if they are querying one or ten XML pathways. This is required for Web 2.0 and
You claim that XPath 2.0 and XQuery being relationally complete can do anything that SQL can do. But, this means relationally complete in SQL operations, not what SQL may be doing inherently. Is full LCA hierarchical processing being performed anywhere else in XPath 02 or XQuery other than being attempted in academic projects? I?ve seen LCA functions being used, but the required LCA logic becomes compounded with the more legs that are involved. With LCA database data use, the LCA?s should remain fixed to be hierarchically database accurate (producing a hierarchically fixed query semantics). Markup processing on the other hand requires variable LCA processing (see VLCA or MLCA) more of a proximity solution, good for full search. This is where XPath?s flexible navigation operations could be utilized; I never said it was not needed in all cases. The separation of Markup from Database data is from my research, but VLCA and MLCA has been researched by others. Check it out with google. The LCA technology may still be academic, except for one important fact, database hierarchical processing with LCA logic is occurring automatically (navigationless), consistently and accurately in ANSI SQL today. The relational/XML data integratin problem was also solved by the natural elevation of ANSI SQL processing to a full hierarchical processing level avoiding the negative effects of the lowest common denominator approach of relational/XML integration today. So Why fight this naturally existing capability ? I believe it is because it is very disruptive to the status quo for SQL/XML and XML product vendors. Customers may having something to say about that.
A XPath/XQuery Challenge:
Can you or anyone reading this email duplicate my simple SQL example in XPath 2.0 or XQuery of joining two nonlinear (multi-leg) structures hierarchically together and applying a multi-leg data filtering? (This is a nonlinear hierarchical filtering, cousin logic). The automatic XML structured formatting can be ignored if you wish (the SQL hierarchical prototype is structure aware enabling this capability). If this example query can?t be performed, can anyone prove why my ANSI SQL syntax and semantics hierarchical prototype example should not work as shown? The example is in my XPath Navigation article, the technology involved is fully described on my site (www.adatinc.com) where the SQL hierarchical processing XML prototype can also be downloaded free. Can any one prove me wrong?
/Mike
Michae M David
Advanced Data Access Technologies, Inc,
www.adatinc.com