Re: Domain selectors

On Mon, Mar 9, 2015 at 12:14 PM, Jens Oliver Meiert <jens@meiert.com> wrote:
>> As an alternative selector syntax for your purpose, a function could be used
>> instead of the attribute-selector syntax you've proposed:
>>
>>     ::domain("example.com") .foo {
>>         /* Domain-specific styles for elements
>>            with the `foo` class. */
>>     }
>>
>> But this does not seem to have serious advantages (other than just
>> somewhat simpler syntax thanks to less braces) over an at-rule like
>> `@document`:
>>
>>     @document domain("example.com") {
>>         .foo {
>>             /* Domain-specific styles for elements
>>                with the `foo` class. */
>>         }
>>     }
>
> I’d prefer ::domain over @document, but I’d wish we pushed harder for
> simplicity. Hence, is that all as simple as we could get it? Why would
> [host=] (or domain, document, or perhaps url) be so objectionable?

What Marat said. Don't try to reuse existing syntaxes for incompatible
things; it's just confusing for everyone.  For the same reason, the
pseudo-element approach is out, unless there's *actually some
pseudo-element in the page that it corresponds to*.

@-rules are how we introduce arbitrary new block-based syntax.  Other
than the exact characters used, there is literally no difference
between "@document domain("foo") { h1 { color: red; } }" and
"[host="foo"] h1 { color: red; }"; the two are expressing identical
semantics, so there's no actual benefit in reaching for the one that
misuses the syntax.

~TJ