Change csv-to-xml() to produce an XHTML table

[michaelhkay]

Change csv-to-xml() to produce an XHTML table

[ndw]

This seems like a very user-friendly change. But I think we should try to produce tables that meet accessibility guidelines. That means scope attributes on column headers and (where rows have headers) row headers.

I'm not sure about that use of title. I think that level of cross-reference would be done with IDs and header attributes in a more accessible table. But generating IDs is fraught with peril.

[ChristianGruen]

I’m not that convinced. Aren’t there much more use cases for processing XML than “just producing HTML”? Would it have been a good idea to introduce an fn:json-to-html function? Some more concerns:

• We have no fixed namespace for the http://www.w3.org/1999/xhtml namespace. Instead of .//fn:row, one would need to write .//*:row or .//Q{http://www.w3.org/1999/xhtml}row.
• For non-HTML usecases, it feels unnatural to work with thead and tbody elements.
• We would reduce consistency (we already have fn:json-to-xml)

With a language that was initially designed for processing XML data, we shouldn’t lose our focus on pure XML processing, and make no concessions for target formats. It would be nice to have something as simple as csv-doc('input.csv')//field[@column = 'City'].