- From: Smylers <Smylers@stripey.com>
- Date: Sun, 4 May 2008 03:16:02 +0100
- To: "public-html@w3.org" <public-html@w3.org>
Daniel Glazman writes: > Smylers wrote: > > > > 3. when I read something like "When the alt attribute is missing, > > > the image represents a key part of the content. Non-visual user > > > agents should apply image analysis heuristics to help the user > > > make sense of the image.", I can't believe my eyes... > > > > Why? That sounds entirely plausible to me. > > Plausible and not implementable. Maybe that's why it's only a 'should'. I'm sceptical as to how successful user-agents will be at divining meaning into unknown images -- but given that this data simply doesn't exist anywhere, a user-agent is in as good a position as anything else in the chain to have a go at it. > A conformance checker will never know what precisely is author's > intent. Yes. I gave examples with <dfn> and <h1> which could be wrong but which a mechnical conformance checker would be unable detect. What else can we do? > And I don't expect non-visual user agents to implement such image > analysis heuristics just because spec authors think they'll do it. Me neither. But even if they don't, people are _still_ going to bulk-upload photos to websites without putting meaningful descriptions on each of them. We could choose to designate all such sites as not being valid HTML 5 -- but that doesn't actually help image-less users to learn what's in the photos. > > I could be persuaded to change my mind by arguments on merit; but merely > > stating that you can't believe why other people don't agree with you > > doesn't really move us on. > > I have no time for rhetorics. Good -- let's stick to technical merits then! > Just the fact the alt section is what it is today - and why it is what > it is today - scares me to death about the whole html5 doc. Reminds > me of the other "kill the style attribute" crazyness. It doesn't scare me. But I don't see how how scared each of us are (but, for what it's worth: please don't die!) helps work out the best thing to do for HTML 5. Smylers
Received on Sunday, 4 May 2008 02:16:49 UTC