- From: Dave Pawson <dave.pawson@gmail.com>
- Date: Sun, 18 Mar 2012 08:43:11 +0000
- To: public-ppl@w3.org
On 18 March 2012 02:38, Arved Sandstrom <asandstrom2@eastlink.ca> wrote: > Hi, Jirka > > With reference to the link you provided for the RenderX validators, yes, > this is what I was looking at too. They'd already had some semblance of > at least one of these way back when. > > What Dave mentioned about grammar-based validation jibes with my > impressions from many years ago; honestly I'd have to revisit the > problem myself before committing an opinion. :-) Having said that, it > strikes me that if you've got folks in Prague at XSL 2012 saying that > XSL validation is a problem, surely a number of them must have tried the > RenderX and AH products, so what are the outstanding deficiencies with > those? There are. Have you looked at it Jirka? For 'the hard ones' they use a content model of text. That's where I came unstuck. This happens in a number of areas. A very good start, but (my view) it's the spec that needs simplification rather than asking the grammar to catch up? border-start-color.attr = attribute border-start-color { text | expr.datatype } expr.datatype = xsd:normalizedString { pattern = ".*\(.*\).*" } A start, but lots more needed for validation using this grammar? I haven't used folint.xsl sufficiently to comment. I could see some combination (schematron like), for the nasty ones? I'd prefer to simplify the requirement, move away from the CSS definition approach to these nasty ones, to provide a grammar for validation. regards -- Dave Pawson XSLT XSL-FO FAQ. Docbook FAQ. http://www.dpawson.co.uk
Received on Sunday, 18 March 2012 08:43:39 UTC