W3C home > Mailing lists > Public > public-css-testsuite@w3.org > February 2013

Re: CSS3-writing-modes: text-combine-horizontal(B

From: Gérard Talbot <css21testsuite@gtalbot.org>
Date: Wed, 20 Feb 2013 23:28:15 -0500
Message-ID: <d316c1b7f3489537d091631081b73a56.squirrel@ed-sh-cp3.entirelydigital.com>
To: "taka" <takaoshiyama@gmail.com>
Cc: "Public CSS test suite mailing list" <public-css-testsuite@w3.org>

Le Dim 17 février 2013 6:34, taka a écrit :
> Hi Gerard
> I respond with this thread as your mail below to be directed here:
> ->
> http://lists.w3.org/Archives/Public/public-css-testsuite/2013Feb/0046.html
> Please review here again, thanks.
> -> http://hg.csswg.org/test/rev/e2ceb488d9f7
>> line 5:
>> <title>CSS3 Writing Modes Test: text-combine-horizontal</title>
>> I suggest
>> <title>CSS Writing Modes Test: text-combine-horizontal - none</title>
>> Nota bene: it should be CSS and not CSS3; we will want to reuse this
>> test if/when there will be a CSS4 writing mode specification.
>> One good method of writing title text in a consistent and predictable
>> manner is:
>> <title>CSS [Name_of_module] Test: [property_name] - [property_value]
>> [optional: some_additional_parameters]</title>
> Thanks, I agree. I corrected each.


please verify again:


and text-combine-horizontal-011.xht
and text-combine-horizontal-012.xht

Another method of writing title text in a consistent and predictable
manner can be:

<title>CSS [Name_of_module] Test: [property_name]: [property_value]
 [optional: some_additional_parameters]</title>

<title>CSS Writing Modes Test: text-combine-horizontal: none</title>

>> -------------
>> line 11:
>> <meta name="flags" content="" />
>> Since your tests are requiring special, specific fonts, do not forget
>> to
>> add the font flag.
> Thanks, I agree. I corrected each.
>> -------------
>> line 12:
>> <meta name="assert" content="This property allows the combination of
>> multiple characters into the space of a single character ." />
>> The assert text describes the goal of the test, what it is testing.
>> One good method of writing the text assert in a consistent and
>> predictable manner is:
>> <meta name="assert" content="This test checks, verifies that
>> [property_name]: [property_value] with [some_additional_parameters]
>> does
>> [expected_results]." />
>> So, here, I would suggest:
>> <meta name="assert" content="This test checks, verifies that
>> 'text-combine-horizontal: none' does not combine multiple characters
>> into the space of a single character." />
> Thanks, I agree. I corrected each.
>> --------------
>> line 39:
>> text-orientaion: mixed;
>> should be
>> text-orientation: mixed;
> Thanks, I corrected typo.
>> -------------
>> line 50:
>> <span class="tcy">1</span>&#26376;<span
>> class="tcy">2</span>&#26085;&#12398;&#27425;&#12398;&#26085;&#12399;&#12289;<br
>> /><span class="tcy">1</span>&#26376;<span class="tcy">3</span>
>> Maybe I'm wrong here but... why is there no span with more than 1
>> character.. Shouldn't that be necessary in the test? Shouldn't that be
>> what would be need to be tested?
> In this test, I address a single digit numeric characters one by one.
> (ideally I would like to handle multiple digits such as '2013' to be
> 'text-combine-horizontal: all', but unfortunately no stable rendering
> techniques for a reference is found.)

Well, then it just means that we can not create a reftest for such test.

The test could still be testing '13' though. When a reftest can not be
created, then it just mean that we can not have a software compare
pixel-by-pixel a screenshot of test with a screenshot of reftest.

A test can be created even though a reftest for it can not be created.
This is where a test can be and will be evaluated as passed or failed
according to a human tester.

One idea would be to reuse the figure 18 example
"Example of horizontal-in-vertical tate-chu-yoko"
from the spec and write a flexible, appropriate pass/fail conditions
sentence for it.

2 Examples given:

>> --------------
>> line 50:
>> class attribute versus id attribute
>> Class attribute is best used for logical grouping of many elements;
>> semantically speaking, this was the genuine design intent for HTML.
>> id attribute is best used for one single element in a document.
>> So, a bit better is
>> <div id="test"> ... </div>
>> <div id="control"> ... </div>
> Thanks, I agree. I corrected each.


<p>Test passes if the upper block is <strong>identical</strong> to the
lower block including <strong>layout</strong> and
<strong>orientation</strong> of each characters.</p>

If you say that 2 things must be identical, then, generally speaking,
people will observe very strictly, very carefully and thoroughly. So,
I'm not sure if "including <strong>layout</strong> and
<strong>orientation</strong> of each characters." is really or
absolutely needed. If you think so, then

<p>Test passes if the upper block is <strong>identical</strong> to the
lower block, including orientation of each character.</p>

The most important word in the test is "identical".

You could also reword the pass/fail test conditions sentence to be saying:

<p>Test passes if there are 2 <strong>identical</strong> rectangles with
identical orientation of characters.</p>

In reftest


line 25 #test
	.tcy  /* tate-cyu-yoko: horizontal-in-vertical */
		text-combine-horizontal: none;
line 31 }
line 32	#control
line 42 <div id="control" lang="ja">...
line 44 <div id="control" lang="ja">...

Remove lines 25 to 31
and replace lines 32, 42 and 44 with

line 32	.control
line 42 <div class="control" lang="ja">
line 44 <div class="control" lang="ja">

Same issue with text-combine-horizontal-011-ref.xht and

line 10:  <meta name="flags" content="font should" />

"should" flag should be used only when a CSS feature is recommended and
not required by the spec. Here, the text-combine-horizontal values are
required. So, here, you can safely remove the "should" flag from
text-combine-horizontal-010.xht and text-combine-horizontal-011.xht (and
probably text-combine-horizontal-012.xht too).

More later

Contributions to the CSS 2.1 test suite:

CSS 2.1 Test suite RC6, March 23rd 2011:

CSS 2.1 test suite harness:

Contributing to to CSS 2.1 test suite:
Received on Thursday, 21 February 2013 04:28:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:13:25 UTC