Simon Pieters wrote:
On Mon, 17 Dec 2007 12:16:17 +0100, Julian Reschke
<[EMAIL PROTECTED]> wrote:
For the record: I totally disagree. It's not the job of the URI and
IRI RFCs to describe how a user agent has to handle things that do not
conform to the RFC3986/3987 syntax.
Why not?
The HTML4 WG had this position about HTML as well. This position is what
makes interoperability suffer, browser vendors having to spend lots of
resources reverse engineering each other, and the resulting de facto
error handling being suboptimal (hard to make extensions, for instance).
Because URIs are just a syntactical construct. If there are common
requirements for *user agents* how to handle things that aren't URIs but
should have been, specify them for these user agents. For instance, why
should RDF or XMLNS care about what *browsers* do with broken identifiers?
BR, Julian