Re: WOFF and extended metadata

On May 27, 2010, at 6:05 PM, Sylvain Galineau wrote:

> Fair enough :)

I see what you are proposing and I understand the syntax. At this point, I'm more interested in what we're trying to solve than I am in doing an immediate up or down vote on your proposal. As I've said, I'm not married to what we have in the spec now (as long as we have a plan for supporting the 1.0 version and, therefore, the files that are already out there) and I am not 100% opposed to your proposal. I just want to have a concrete "this is why we are doing it" before I say, "Hey, that's new! Let's do it!"

As best I can gather, the concern right now is that the XML is too open-ended and could lead to problems in UAs. That is a fair point. By adding a specific block and structure for WOFF makers to put things into, you make an attempt to prevent a UA from getting arbitrary, open-ended XML. My point, and one that you mentioned as well, is that it doesn't stop arbitrary XML from appearing outside of the metadata-extended block. So, I don't see how metadata-extended is completely solving this problem.

My proposal was to define the limits of what a UA can expect in the metadata structure. This would define a point at which a UA has the right to stop reading/rendering data.

>> Okay, so your code would support this abstraction:
> 
> Not quite. I don't allow arbitrary XML.

Sorry, I should have given more context than the snippet from your email. What I meant was, if we approve your proposal, how would your "write it once and never bother with it again" UI code in IE handle this?

<metadata>
	<blahblah>
		<foo>
			<bar>
				<lalala/>
			</bar>
		</foo>
	</blahblah>
</metadata>

Will you ignore the blahblah element because it is not in the spec? What if it is part of a new version of the spec that was released after your code was written and the browser shipped? Show it? Discard it? If your answer is "it will ignore the elements that are not part of the spec", how does your code know what is in the spec and what isn't? Do you update it each time the spec is bumped? Isn't that what we're trying to avoid?

Your comment about writing the UI code once and not touching it again was illuminating. It gave me a good idea of what browser makers could want. I'm trying to align that with what the spec should say about the handling of arbitrary, open-ended XML in the metadata.

Tal

Received on Friday, 28 May 2010 00:52:03 UTC