RE: RDFa and Web Directions North 2009

Julian Reschke:
> The implementation cost of these attributes is zero: recipients are not
> required to do anything with them.
> 
> Documenting them (in tutorials, books...) is not free, but then, if you
> aren't interested in them, why document them?
> 

Any feature of any computer system has potential costs in terms of confusion, training etc. Even things you don't use, and can ignore. There is a cost to deciding to ignore something.

When I buy an HTML5 book I expect it to document the whole language: the language necessarily includes all attributes, but perhaps not all script types. Thinking about how I would navigate around such a book, I suspect it would be easier to not visit RDFa material if there were no related attributes.

[I, of course, would want to visit the RDFa material, but in a world which expects to see the value of RDFa in practice before having to read up on it, we perhaps need to put ourselves in the mindset of someone who doesn't]

> Are you suggesting that only putting the attributes on <script> somehow
> changes costs? How so?
> 

That is quite a good question.
& I am far from clear that I have a good answer.

The base of users who bears the cost are those HTML authors and other users who might view source.

Attributes, particularly generic attributes that can go on many elements, appear in an (informal) namespace that when I am viewing source I quite want to get my head around.

If I see a script element with a type that I've never heard of, I think I can more easily skip over it.

If I run tidy on some html I expect to be told about unexpected attributes ... but not so much about unrecognized script types.

Jeremy

Received on Sunday, 15 February 2009 02:18:45 UTC