Re: [css-counter-styles] Allow impls to extend range of hebrew

On 05/01/2015 04:26 PM, fantasai wrote:
> On 04/21/2015 01:44 AM, Richard Ishida wrote:
>> On 21/04/2015 08:56, Simon Montagu wrote:
>>>> I bring this up because I see a testcase which tests 20000/20001 fails
>>>> >on our impl [2], but AFAICS, the behavior of WebKit/Blink matches ours
>>>> >(except that they don't reorder the marker text correctly.) And I
>>>> think
>>>> >that test should be rejected.
>>>> >
>>>> >[1]http://dev.w3.org/csswg/css-counter-styles/#hebrew
>>>> >[2]
>>>> >http://www.w3.org/International/tests/repo/results/predefined-counter-styles#simplenumeric
>>>>
>>>> >
>>>> >- Xidorn
>>> I agree that the implication in the test that Hebrew counter style must
>>> only be implemented up to a fixed ceiling is problematical.
>>
>>
>> the test currently reflects what the spec says. If the spec is changed,
>> then the test can be changed too.
> 
> The CSSWG resolved to allow implementations to implement past the ceiling
> (but must implement up to the ceiling, of course, and must use the fallback
> style for whatever happens after the UA's own ceiling).
> 
> So that means the test needs to allow two different renderings: one with
> the fallback, one with a correct Hebrew numeric representation. (CSSWG
> reftests can express this requirement by linking to two separate
> references.)
> 
> 
> Simon, is there any number that is definitively past the limits of the
> numbering system?

No, the numbering system just gets more and more unwieldy, and more
liable to ambiguity because of the lack of a zero symbol, as the numbers
get higher. Its a judgment call at what exact point it becomes unusable.

Received on Sunday, 3 May 2015 08:01:36 UTC