Re: [css-variables] How to spec the OM for vars?

On Sat, Aug 25, 2012 at 1:12 AM, Tab Atkins Jr. <jackalmage@gmail.com>wrote:

> On Thu, Aug 23, 2012 at 10:04 PM, Glenn Adams <glenn@skynav.com> wrote:
> > On Fri, Aug 24, 2012 at 12:55 PM, Boris Zbarsky <bzbarsky@mit.edu>
> wrote:
> >>
> >> On 8/23/12 6:22 PM, Tab Atkins Jr. wrote:
> >>>
> >>> That works nicely with the existing attributes, I assume?  Or do I
> >>> need to write the getter/setter to handle them as well?
> >>
> >> So my personal preference would be to use the IDL "getter" and "setter"
> >> syntax for vars only and use normal IDL attributes for actual
> properties.
> >
> >
> > My concern on this is that it will be a rather never-ending spec
> maintenance
> > problem, requiring new partial interfaces on a regular basis (i.e., every
> > time new CSS props are introduced or changed).
>
> This is fine.  Each spec that adds new properties just puts a little
> IDL section with this in it:
>
> partial interface CSSStyleDeclaration {
>   attribute DOMString? newProperty;
> }
> Amend the table at ... to contain the following additional rows:
> | newProperty | new-property |
>
> It's pretty trivial.
>
>
> An alternative is just to hack around WebIDL here, and simply state
> that the interface has a number of additional attributes of this form,
> which map (as specified) to the CSS property obtained by transforming
> the name by <specify camelCase to dash-separated>.
>

Which would need to be qualified with "If and only if a UA semantically
supports property x, then it shall implement "partial interface
CSSStyleDeclaration { DOMString x; }" etc, since as we have seen above, UAs
are presently not implementing attributes for unsupported properties. Or
would you mandate they support the attribute in any case, returning null if
not semantically supported?