Re: [eXist-TEIXML] passing arguments to XQuery functions

[Christopher T. <chr...@ca...>]

Hi Joe,

Thanks for the advice about namespaces - I will keep your 
troubleshooting statistics in mind as I continue with this project.  If 
all goes well hopefully I will have a working website by the time that 
anniversary comes around :)

Cheers,
Chris

On 18/03/12 6:04 AM, Joe Wicentowski wrote:
> Hi Chris,
>
>> Thanks again for your help on this, and to Joe for setting up this list for newbies like me.
> I'm so happy to hear that you, together with Martin and Peter, have
> found a solution!  Since I didn't participate in the discussion (work
> busy-ness + a bad cold), let me offer some reflections:
>
> First, Chris, you're graduating to a really exciting level of XQuery
> development sophistication.  You're putting commonly used application
> logic into functions and moving those functions into library modules.
> This isolates the code, centralizes the functions, and makes the code
> more portable, more reusable, and much easier to test.  Your function
> takes an entire TEI document as an argument, and thus it can be used
> on any TEI document in your database, now or as you add new documents.
>   If your TEI document structure changes in the future, you only have
> to modify this one function to ensure your application runs as before.
>
> Second, namespaces were central to resolving the problem.  My rule is
> that if I write a query and get nothing instead of the expected
> result, namespace problems are the cause of the problem 90% of the
> time.  If I check the XPath and the assumptions in the logic, and
> everything still looks right, that 90% rises to 99%.  This isn't to
> say that namespaces are evil; they're essential, but require
> annoyingly absolute precision.
>
> Third, default namespaces.  Declaring a default element namespace can
> be very handy, particularly if all inputs and outputs are in the same
> namespace and no elements are in the empty namespace.  But in my
> experience, I often find myself removing such declarations.  Declaring
> TEI or HTML as the default makes it difficult to address empty
> namespaces ( *:foo is a kludge and not good for performance), and in
> my applications I almost always mix TEI, HTML, and empty namespaced
> nodes.  But absolutely, do try it out and keep this technique in your
> arsenal, but I personally find it of use in marginal circumstances.
>
> Finally, this list will have its first anniversary in a few months!
> It'll be an opportunity for us to reflect on what we've learned this
> year, and on what we can do looking ahead.
>
> Thanks and best wishes to you all,
> Joe