Re: [CSS21] Multiple IDs

On Thursday, August 25, 2005, 5:55:04 PM, Malcolm wrote:

MR> On Thu, Aug 25, 2005 at 03:41:51PM +0200, Chris Lilley wrote:
>> Removal of the wording on multiple ID attributes limits the suitability
>> of CSS 2.1 to SVG and CDF. SVG and CDF WGs request that this
>> functionality remain in the draft and be tested for in the test suite.

MR> Firstly, the suggestion is to downgrade this note from normative to
MR> informative, in order to permit CSS2.1 to exit CR, not to remove the
MR> feature entirely.

Which would then mean it could not be tested for conformance and would
not feature in the test suite.

>> In SVG 1.2, IDs could arise from any of these methods. Why should
>> CSS selectors give different results to getElementById? Why does CSS 2.1
>> propose to tolerate this variability and inconsistency, whereby
>> compliant implementations can fail on a simple #foo selector?

MR> Secondly, I don't believe that anyone is suggestion that UAs that
MR> support multiple IDs in other ways (via getElementById, for example)
MR> should _not_ support multiple IDs via the CSS ID selector, it's just
MR> that those UAs that don't support it in CSS don't support it
MR> anywhere else either.

The variability concerns me, and the fact that we might be stuck with
that variability for a very long time.

MR> But most importantly, shouldn't your comment really be directed at CSS
MR> implementors rather than the CSSWG?

If the CSS WG makes something informative and optional, the implementors
will not implement it anyhow out of the goodness of their hearts, no.
Well maybe in some cases they will - but if its required and its tested
in the test suite they are much more likely to.

MR>  It makes no sense to request the
MR> working group leave the feature in if there are no interoperable
MR> implementations, as CSS 2.1 won't exit CR until that happens.

Ah, I see what you are saying. Yes, implementors please realise that
with CDI coming up, this feature becomes more important.

MR> [Alternatively, if you're already aware of two interoperable
MR> implementations of this functionality, you don't need to do
MR> anything, as the feature will only be dropped if two implementations
MR> cannot be demonstrated at the end of the CR stage.]

It will be much easier to test whether two implementations are
interoperable once there are tests on this feature to test them with.

-- 
 Chris Lilley                    mailto:chris@w3.org
 Chair, W3C SVG Working Group
 W3C Graphics Activity Lead

Received on Tuesday, 30 August 2005 01:53:47 UTC