Re: [css3-images] Features Overview

On 05/06/2011 02:07 PM, Boris Zbarsky wrote:
> On 5/6/11 4:52 PM, fantasai wrote:
>>> Works for me in Gecko, WebKit, Presto, IE9.
>>
>> We would presume that because SVG requires an identifier for id attributes.
>> Your example doesn't validate.
>
> Which doesn't mean that no one is using it.

True. But it makes it less likely.

>>> Basically, I think the media fragments draft is not
>>> backwards-compatible with current behavior and thus I think that using it
>>> should require explicit opt-in.
>>
>> Your argument assumes that there are enough authors who *hand-write* their
>> SVG identifiers *such that* they are invalid in a way that matches #xywh
>> syntax, that this is a problem.
>
> Or machine-generate such that they are invalid. Since it works in all
> browsers, would they even notice?
>
> Also note that "enough" doesn't have to be many to cause compat issues.
> Just one is enough.

One machine generator that generates xywh notation, yes. But given
the awkwardness of the media fragment syntax wrt use as an identifier
and the fact that it's invalid, I'm reasonably confident that there
isn't such a machine generator.

If we were talking about numbers as identifiers, or putting spaces in
identifiers, or using a non-identifier punctuation character like ':',
that would be a more reasonable concern. But we're talking about
something that matches "xyzh=\d+,\d+,\d+,\d+".

> So the only question is how much hoop-jumping is worth it to avoid the
> compat issues. You're making the argument that 0 is the right amount, yes?

It depends on the magnitude of the compat problem. I believe the
problem you're concerned about is entirely theoretical.

~fantasai

Received on Friday, 6 May 2011 21:42:32 UTC