[xsl] collection() and uncommon file extensions

["Martin Holmes gtxxgm-xsl-list-2@xxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

Hi all,

The recent changes to XPath 
(https://www.w3.org/TR/xpath-functions-31/#func-collection) have 
introduced the capability for the collection() function to retrieve 
non-XML documents as well as XML documents. However, that has broken 
some processes I have where XML documents with different extensions are 
being retrieved. For instance, where this:

collection('dir/?*.hocr')

used to happily retrieve and parse HOCR files (which are actually 
XHTML), Saxon now treats these files as xs:base64Binary items, and won't 
parse them, even though they have XML declarations.

I know that the recommended approach to dealing with this is to use a 
Saxon configuration file to register the file extension -- which I 
presume would be done like this:

<resources>
  <fileExtension extension="hocr" mediaType="text/xml"/>
</resources>

However, this doesn't seem to work for me -- do I have that syntax wrong?

Also, the conf file approach isn't easily portable, so I'm wondering if 
there are any plans to enable the media type to be specified on the 
collection() function itself, or to be registered in an XSLT document 
somehow?

Cheers,
Martin