Re: Armenian tests for CSS 2.1

On 09/10/2010 08:28 AM, Richard Ishida wrote:
> These tests were actually largely exploratory in the i18n test suite,
> and aimed at CSS3.  I'm thinking that they should probably be changed
> for use with the CSS 2.1 test suite.  Currently CSS 2.1 says next to
> nothing about Armenian numbering except, in the errata[1], that
> 'armenian' means uppercase Armenian.
>
> For CSS 2.1 it seems to me that the only thing we can really test is
> whether using the value 'armenian' produces some Armenian numbering,
> and that that numbering is uppercase.
>
> The maximum number is undefined. Note that I recommended [2] that
> CSS3 only specify up to 9,999 in CSS3 because of difficulties in
> ascertaining how to handle Armenian numbers after that. On that
> basis, I think we should discard the test at
> http://test.csswg.org/suites/css2.1/20100815/html4/list-style-type-armenian-002.htm.

Done.

> Note also that there is also some ambiguity around the 700s (see [2]),
> but although IE followed my recommendation on that, we don't have any
> text in the spec related to it, so we should not test for it.
>
> I therefore suggest that for CSS 2.1 tests we reduce
> http://test.csswg.org/suites/css2.1/20100815/html4/list-style-type-armenian-001.htm
> to a maximum number of, say,  100, or less, and discard test 2.  And
> that test one check for uppercase numbering specifically.

Done.

~fantasai