- From: David Carlisle <davidc@nag.co.uk>
- Date: Mon, 7 Jan 2008 15:43:31 GMT
- To: hsivonen@iki.fi
- CC: www-math@w3.org
> Note that I was talking about document conformance to err on the > prudent side when it comes to actual interop issues. I was silent on > whether processing in UAs should be changed not to trim whitespace in > attribute values. Ah. Yes being stricter in what you generate than what you accept is of course sensible. Currently though we just structure the definition (of these attributes and everything else) with a single grammar, which would need to be the more permissive one as a grammar is (by definition at least) the definition of what a system should accept although of course it can also be used to give hints to document generation tools. There could of course be a separate strict grammar or prose text giving a conformance requirement for authoring systems, if that's what we decided we wanted.... I think in most schema languages specifying an attribute to take one of an enumerated list of tokens will by default imply white space stripping (as that is what the XML spec specifies for for any attribute type other than CDATA and xsd specifies for every primitive type except xsd:string) but of course the stricter version could be done by restricting xsd:string with a regular expression pattern. David
Received on Monday, 7 January 2008 15:44:08 UTC