Re: IGNORE CASE IN XSLT

Yes, quite true, and certainly the case of a super-functional tool (an XML parser plus other features) is an interesting one to consider when trying to define "conformance".

A strict definition of conformance that says "what the XML Rec describes, and no more" is important to maintain if only so we avoid the situation where, say, BigSoft Corporation releases an "XML processor" that supports non-standard features, such as (say) SGML tag minimization. Then the market starts working with tag-minimized non-XML. If a significant portion of the market then locks into this, BigSoft's competitors have to match this feature, and pretty soon it's not at all clear that tag minimization is non-standard.

Note this phenomenon is not less dangerous, but rather more so, when the feature is one that the market (or part of the market) really wants. That is, it has nothing directly to do with the technical merits.

The good side of this is that it represents, arguably, evolutionary pressure on the standards to give the market what it wants. The bad side is that it may put BigSoft in the position of dictating what the actual standard is.

This is why standards hawks such as myself are so guarded about non-standard features. Labelling them as such, by saying "this includes an XML processor but has some extensions and non-standard features" helps a lot.

Reconsidering my remarks, I don't think the ignore-case option is sufficient. Too often SGML and HTML are not well-formed. One indeed would be better off with an SGML parser.

It depends on the case, as you'll surely agree. In this case, IIRC the OP wanted to map "italic" and "ITALIC" to "i" (not simply case-folding), but didn't have to worry about well-formedness.