I've noticed that draft-ietf-atompub-autodiscovery-01.txt doesn't mention
XML namespaces and tag prefixes. XHTML can get even more complex than memo
suggests:
<foo:link xmlns:foo="http://www.w3.org/1999/xhtml"; rel="alternate"
type="application/atom+xml" href="bar"></foo:link>
My suggestion is that instead of specifying all quirks of X[HT]ML syntax:
* require that XML parser is used for XHTML
* if document turns out not to be well-formed (which often is the case
with real-world "XHTML"), allow HTML parsing rules used as fallback
And then simply state that in XHTML <link> element in
"http://www.w3.org/1999/xhtml"; namespace should be used. An example XPath
expression or W3C DOM-based pseudocode might be very helpful there.
BTW: in all examples attributes have always the same order. They could be
shuffled to emphasise that order is not significant.
--
regards, Kornel Lesiński
