- From: Norm Tovey-Walsh <norm@saxonica.com>
- Date: Tue, 12 Nov 2024 08:12:59 +0000
- To: Steven Pemberton <steven.pemberton@cwi.nl>
- Cc: ixml <public-ixml@w3.org>
- Message-ID: <m2cyj0kgz8.fsf@saxonica.com>
Steven Pemberton <steven.pemberton@cwi.nl> writes: > Enclosed is a proposal, or technique, depending on your point of view, for including namespaces in serialisation of ixml. It will work in many existing ixml implementations. Note that it uses a feature not yet in 'standard' ixml, but already implemented in several implementations. Well, it’s clever, but I really actively dislike it. It treats namespace declarations as if they were attributes, which they are not. Let’s not pretend this is 1998 and there are still mainstream XML parsers that are not namespace aware and actually put namespace declarations in the data model as attributes. The resulting grammar is difficult to read. So difficult to read that I begin to think we should reconsider allowing aliases to be strings. Using rewriting, it would be easy to construct XML that’s not namespace well-formed. We must call that an error, so I think even iXML processors that generate textual XML output will have to do the work and determine if the result is namespace well formed. If users want namespaces (users want namespaces) and we are ready to consider adding support for them to iXML (maybe we are, but we’ve been resistant in the past), let’s think it through and provide a mechanism that doesn’t rely on weird lexical rewriting tricks. That way lies confusion, error, and madness. Be seeing you, norm -- Norm Tovey-Walsh Saxonica
Received on Tuesday, 12 November 2024 08:13:06 UTC