- From: Tony Ross <tross@microsoft.com>
- Date: Tue, 4 May 2010 00:16:45 +0000
- To: Jonas Sicking <jonas@sicking.cc>
- CC: "Ennals, Robert (robert.ennals@intel.com)" <robert.ennals@intel.com>, "public-html@w3.org" <public-html@w3.org>
On Monday, May 03, 2010 4:43 PM, Jonas Sicking wrote: > On Mon, May 3, 2010 at 4:20 PM, Tony Ross <tross@microsoft.com> wrote: > > > > I agree with not wanting to overhaul the language. > > Really I'm just looking to let something like the following validate: > > > > <th asp-sort="desc"> > > > > This should require zero changes to the parsing algorithm and resulting DOM. > > Unless I misunderstood, this is what Rob's proposal Y enables. > > I guess I was a bit over dramatic with "overhaul the language" :). It doesn't > mean any changes to implementations, but it does change a lot of other > aspects. It does change what is "valid HTML" (though the usefulness of that > term is debatable), and it does change logistics in that now HTML is run > through a registry in addition to a working group. And it also creates the > problem you mentioned earlier of people rushing to register short prefixes, > not to mention that we'll probably end up loosing good names to defunct > projects and companies. What do you think about supporting the syntax without requiring the registry? > > The distinction is between attributes defined by a library author and > > attributes defined by the page author. > > Though arguably the data-foo-bar vs. data-bar distinction does that already. > At least if people play nice. And if people don't play nice there is nothing > preventing page authors using attributes named asp-backgroundcolor. True, but at least they had to choose a prefix with the non-"data-" scheme. And they potentially ended up with a shorter attribute name at the same time. With "data-" alone, trying to keep names short makes "no prefix" highly desirable. > > > > I'm fine with putting the burden of conflict-resolution on the library itself. > > > > Perhaps we can add text encouraging libraries to make their > > > > prefixes configurable. > > > > > > That sounds like a good idea to me. > > > > > > > Great. Does anyone disagree with this? > > If not, I think this point would be good to include in Rob's proposal Y. > > I think we should do this apart from any of the decentralized extensibility > issue entirely. The spec already recommends libraries on how they should > use data attributes. Seems like it could go in as part of that. I'm ok with that so long as other proposals preserve the recommendation. -Tony
Received on Tuesday, 4 May 2010 00:17:19 UTC