W3C home > Mailing lists > Public > public-css-testsuite@w3.org > December 2015

Re: [css-writing-modes-3] text-orientation-mixed-001 , text-orientation-010 , text-orientation-014 tests

From: Gérard Talbot <css21testsuite@gtalbot.org>
Date: Sat, 12 Dec 2015 02:40:33 -0500
To: Koji Ishii <kojiishi@gmail.com>, Kazuaki Takemura <takemura@networksoft.co.jp>
Cc: Public CSS Test suite mailing list <public-css-testsuite@w3.org>
Message-ID: <d1b5fd3c97055d19e9e1edae4d48a826@gtalbot.org>
Le 2015-12-07 10:26, Koji Ishii a écrit :
> 2015-12-04 7:43 GMT+09:00 Gérard Talbot <css21testsuite@gtalbot.org>:
>> Hello,
>> http://test.csswg.org/source/css-writing-modes-3/text-orientation-mixed-001.xht
>> The reference image must be redone so that it uses a "T" (65332) and 
>> not a
>> "T" (84).
>> More info:
>> http://www.gtalbot.org/BrowserBugsSection/CSS3WritingModes/T-in-text-orientation-mixed-001-test.html
> I think it's intentional and correct, though I agree that it's not easy 
> to
> understand. "T" is in FULL WIDTH T to set in upright. If you change it 
> to
> ASCII T, it must set in sideways, and the test is no longer testing 
> "mixed"
> value.

Kazuaki Takemura,

I have updated your


test a bit:


I wanted to approve your text-orientation-mixed-001 but "if both "Text 
sample" rectangles are identical" can lead testers to believe that 
everything must be pixel-perfectly-identical. I think we need to 
rephrase the pass-fail-conditions sentence to just say something like:

Test passes if both "Text sample" rectangles have only their "T" 
displayed vertically and their "ext Sample" displayed sideways (90 
degrees clockwise).

- - - - - - -

>> http://test.csswg.org/source/css-writing-modes-3/text-orientation-010.xht
>> The second (bottom) part of the test is wrong, not best, not testing 
>> what
>> it claims to be testing because #test and #control use identical code. 
>> The
>> initial, default value of 'text-combine-upright' is 'none'; the 
>> initial,
>> default value of 'text-orientation' is 'mixed'.
> Huh, agree that this test does not make sense, it only makes sense to 
> who
> understands how it should look, and test and ref are not related.

Kazuaki Takemura,

I am trying to rehabilitate your text-orientation-010 test as much as 
I wanted to use an image as reference (instead of, in place of your 
current code #control which can not be correct) but Chrome 49 uses a 
large letter-spacing (or inter-character spacing) which is larger than 
Firefox 45. This may be allowed by the spec... I do not know for sure. 
So, I need to think this over some more.

- - - - - - -

>> http://test.csswg.org/source/css-writing-modes-3/text-orientation-014.xht
>> The second (bottom) part of the test is wrong, incorrect. The initial,
>> default value of 'text-combine-upright' is 'none'; the initial, 
>> default
>> value of 'text-orientation' is 'mixed'. So, 'sideways' text and 
>> 'mixed'
>> text are compared.
> This test is not good, but does not look wrong to me.

Kazuaki Takemura,

I am also trying to rehabilitate your text-orientation-014 test as much 
as possible.
I am thinking of using

	font-family: "DejaVuSerifBook";
	src: url("support/DejaVuSerif-webfont.woff") format("woff");
	/* Filesize: 18096 bytes (17.7 KBytes) */

since all the characters are basic latin and then create an image as 
reference for comparison, for control. In your current test, #control 
can not - in all fairness - serve as a reference for comparison.

Overall, I am more and more enclined to believe that we should split 
your text-orientation-010, text-orientation-011 and text-orientation-014 
tests in 2 distinct, separate tests so that some tests would say 
"identical", others would grant some latitude, leniency with 
inter-character spacing.

Test Format Guidelines

Test Style Guidelines

Test Templates

CSS Naming Guidelines

Test Review Checklist

CSS Metadata
Received on Saturday, 12 December 2015 07:41:05 UTC

This archive was generated by hypermail 2.4.0 : Friday, 20 January 2023 19:58:21 UTC