Re: [xsl] design question on function namespaces

[Bruce D'Arcus <bdarcus@xxxxxxxxx>]

On May 3, 2005, at 8:58 AM, Michael Kay wrote:

> The thinking here is that it's often useful to create a library of 
> reusable
> functions that compute derived data from your source document: for 
> example,
> "does this chapter have any footnotes" or "how many columns are there 
> in
> this table" or "is this employee in the pension scheme" or "get all the
> products that this customer has bought". You should think of these as
> methods associated with the element in question (even though there's no
> encapsulation or inheritance), and recognize the link between code and 
> data
> by matching up the namespace.

Yes, I understand this.  But say you want to implement a function for 
your "does this chapter have any footnotes" example, but you deal with 
DocBook NG, TEI 5 and OpenOffice.  Would you reimplement the function 
three times, or would there be a way to make it generic so that it 
could be reused across the different document types and namespaces.  If 
you do the former, then they're not THAT reusable.

My thinking here is partly conditioned by learning more about the SRU 
and SRW search and retrieve protocols, and their CQL query language, 
all of which come out of the library world.  There you have abstract 
queries defined in indices.  So, if you query the Library of Congress 
SRU (RESTful) server, you don't do 
"mods:titleInfo/mods:title="whatever"; you do "title=whatever" and let 
the server determine what to do with that.

Example:

http://z3950.loc.gov:7090/Voyager?
operation=searchRetrieve&version=1.1&query=dc.title=XSLT&recordSchema=mo
ds&startRecord=1&maximumRecords=20

I know XSLT presents limitations to implement this sort of thing, but I 
guess I'm thinking about the possibility of a similar sort of abstract 
functions -- like "bib:getYearIssued" -- that could be extended to take 
different input sources.

Bruce