[xml-dev] The limitations of XPath and navigation for XML database proce

["Ken North" <kennorth@sbcglobal.net>]

>> If he's going to claim that XPath is procedural when the rest of the world
>> claims it is declarative, then at least he ought to explain why he holds
>> this view

For a number of years, Mike worked on the NCR Teradata product. It's a DBMS 
targeted at the large data warehouse market. So he comes at the declarative vs. 
navigational issue from the perspective of working on a platform where there was 
an SQL engine and query optimizer. Declarative (non-procedural) is the 
alternative to navigational or positional logic.

To someone building SQL platforms, the distinction is something like Chris 
Date's book title "What Not How". The goal of SQL was to provide a language that 
enabled a user to specify what query results were desired, without specifying 
how to navigate to the target data. It's the job of the SQL optimizer to 
determine the best path for accessing the data, not the user who writes the 
query.

The analogy I've used to explain this is contrasting how people use a taxi. The 
passenger who says "Take me to the airport" and leaves it to the taxi driver to 
determine the best route is like the SQL user who leaves it to the optimizer to 
determine the best access plan.

A different passenger might tell the taxi driver, "Take Interstate 15 to Highway 
78 and turn west. Then take Interstate 5 South to the San Diego Airport exit". 
That's navigational, not declarative.

So the question of whether a query engine supports declarative or navigational 
queries is essentially whether the user needs to know the physical structure of 
the data and how to navigate to it.