Re: Review of 'row-progression-01[1-8].xht'

Le 2015-01-11 19:02, Gérard Talbot a écrit :
> Le 2015-01-11 03:22, 塩澤 元 a écrit :
>> Gérard,
>> I have submitted 'direciton: rtl' version.
>> Please review these test-cases.
> Hajime,
> I think those tests should not be using 'unicode-bidi'... unless you
> want to specifically test directionality of text inside table cells.
> I think your tests - without use of 'unicode-bidi' - should be about
> *_table column ordering_*; how 'direction:rtl' set on a table element
> affects table column order.
> When you read the spec and a testable statement of the spec, you want
> to create a test that will check one and one single aspect of the spec
> and one and only one statement of the spec. And you want to create
> filenames for test and test assertion text that will be coherent.
> When I now rethink this, I do not see how we could create more tests
> or other tests regarding table rows progression. By definition,
> 'direction: rtl' and 'unicode-bidi: bidi-override' are not supposed to
> have any impact, any rendering effect on table row progression to
> begin with! But 'direction: rtl' should have an impact, a rendering
> effect on columns order.

I have re-read the spec again and checked a bunch of 'direction' and 
'unicode-bidi' tests we did in CSS2.1 test suite

and clearly this is the case.

Only writing-mode has a rendering effect on row progression (or, if you 
prefer, row right-to-left order for 'writing-mode: vertical-rl' tables 
or row left-to-right order for 'writing-mode: vertical-lr').

2- Setting 'direction' on a table element only has a rendering effect on 
columns order (or, if you prefer, column top-to-bottom order or column 
bottom-top order)

3- setting direction on a block container sets the text direction but 
bytes and characters are not reordered (or bidi-reordered); you need 
'unicode-bidi: bidi-override' to do that.

> Are your 'row-progression-01[1-8].xht' tests a waste of time? No. They
> should probably a) be filename-renamed b) with new <title> and c) with
> new text assert and check how 'unicode-bidi: override' does to text
> directionality in table cells and another version of those same
> 'row-progression-01[1-8].xht' tests should be about table column
> order.
> I'll try to create an example of this later.

Hajime, here's the example I had in mind. I took your

and then created 2 distinct, modified variants of it:


Chrome 39 must have a bug with the rendering of "H&nbsp;" and "V&nbsp;" 
in s21direction-table-column-order-011-Hajime.xht test because it fails, 
if I'm correct, to start the 4th line box with the H (or V) instead of 
the non-breaking-space.

IE11 fails completely s21direction-table-column-order-011-Hajime.xht

IE11 does not, it seems, support 'unicode-bidi: bidi-override' for 
vertical-rl tables.

Test Format Guidelines

Test Style Guidelines

Test Templates

CSS Naming Guidelines

Test Review Checklist

CSS Metadata

Received on Monday, 12 January 2015 04:19:36 UTC