- From: Simon Pieters <simonp@opera.com>
- Date: Fri, 30 Jan 2009 13:22:03 +0100
- To: "David Singer" <singer@apple.com>, "Michael(tm) Smith" <mike@w3.org>
- Cc: "Ian Hickson" <ian@hixie.ch>, public-html <public-html@w3.org>
On Fri, 30 Jan 2009 11:38:12 +0100, David Singer <singer@apple.com> wrote: >> It's certainly the case that the HTML5 draft doesn't confine its >> definition of what a conformant document is to only what's >> machine-checkable. I think that's a good thing, and I think >> anything else that sets out to describe what a conformant document >> is should also not confine itself to only what's machine >> checkable. On the fact of it at least, it does seem to me that >> "document doesn't contain relative URLs when the base URL can't be >> used to resolve URLs" seems like a constraint that ought to be >> described in my draft. > > I appreciate it's more than a syntax question; that's what makes it > perhaps not so naive. I was pondering the whole question "what if the > URL is not capable of being de-referenced?" (e.g. > http://deliberately.unknown.host.xw/). Then there is the question of > URLs with unknown or inappropriate methods...clearly <something > src="mailto:someone@w3.org" /> is pretty odd, and one would be tempted > to say that the URL here must be a form that delivers content. But then > is <something src="daveprotocol:random.content" /> conforming or not, if > you don't know what daveprotocol does...? I think it comes down to the question of what the concept document conformance is for, which is for a specific QA tool, namely a validator. The idea is to set out some rules so authors can catch their mistakes and get out of bad habits. In the case of URLs, the spec currently has a concept of "valid URL" to catch violations against the IRI spec and to avoid weird behavior with legacy encodings, but for a QA tool to be useful to catch typos and mistakes in URLs what one further needs is a link checker that tries to follow the links and reports 404s, etc. I don't think we should say that, for instance, linking to a 404 resource is non-compliant, so even if we expand on the rules for document conformance around URLs a validator still wouldn't be very useful for catching URL mistakes. -- Simon Pieters Opera Software
Received on Friday, 30 January 2009 12:22:56 UTC