Re: Help wanted for CSS Test Suites and WebKit implementations

Le Lun 21 janvier 2013 4:02, taka oshiyama a écrit :
> Hello,
> I plan to resume CSS test suites re-submissions against my previous ones
> which were submitted about two years ago.
> I hope your and other people's assistance for further review and
> approval.
> regards,



Taka,

I examined many tests in
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/

Here's a few reviewing comments.

1-
In many tests, replacing "square box" with "rectangle" is a bit more clear.
Many tests use this typical pass/fail conditions sentence:
"Test passes if each pair of upper and lower text in the square box is
identical."
when I believe this wording could be improved and be a bit more clear. I
do not have a good, sure and safe proposal at this time. I think many
tests should be broken into smaller sub-tests.


2-
In all reftests (ending with "-ref.xht"), the title should be
<title>CSS Reftest Reference</title>
http://wiki.csswg.org/test/reftest#the-reftest-reference-file

Eg.
<title>CSS Test: hanging-punctuation - allow-end - basic cases Reftest
Reference</title>
in
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/reftest/hanging-punctuation-allow-end-001-ref.xht
should be replaced with
<title>CSS Reftest Reference</title>
If you still want to include such info, then you could write it as a
<!-- comment -->

<title>CSS Test: Reftest Reference</title>
<!--
Reftest for hanging-punctuation - allow-end - basic cases
-->

Shepherd Test Manager will keep a track, a record of all tests in
relation with their respective reftests via the "Referenced by" field.
Eg.
http://test.csswg.org/shepherd/reference/hanging-punctuation-allow-end-001-ref/

3-
xml:lang="ja" and content-language. I believe these tests should be
declaring "ja" and "en" but I am not, at this time, perfectly sure how
this can be done inside those tests. I propose

<meta http-equiv="Content-Language" content="en, ja" />
<!-- http://www.ietf.org/rfc/rfc2616.txt, section 14.12 -->
"Multiple languages MAY be listed for content (...)"
which meet exactly these tests' conditions.


4-
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/line-break-normal-001.xht

http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/line-break-strict-001.xht

[
Test passes if the highlighted character in the first line is located at
the same position in the next line.
* This is a recommended part ("should").
]

[* This is a recommended part ("should").] is really a technical issue
that normal testers should not be notified of. The only important issue
is that the "should" flag is listed in the <meta name="flags"> of the
test.

Maybe this test should be broken into smaller tests so that the wording
of the pass/fail conditions would be easier to edit and easier to
understand:
"Test passes if the highlighted character in both rectangles is at the
exact same horizontal position."

5-
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/text-decoration-line-overline-001.xht

<link rel="help" title="CSS Text Level 3: 11.1.1. Indent Text: the
'text-decoration-line' property"
href="http://www.w3.org/TR/css3-text/#text-decoration-line" />
(...)
		<meta name="flags" content="ahem" />
(...)
		<style type="text/css"><![CDATA[
				#reference
				{
					border-top: solid 1px black;
				}
				#test
				{
					text-decoration-line: overline;
				}
		]]></style>
	</head>
	<body>
		<p>Test passes if one line appears above the "Filler Text".</p>
		<div id="test">Filler Text</div>
		<div><br/><span id="reference">Filler Text</span></div>

The pass/fail conditions sentence is confusing. If testers would
strictly follow what it says, then all browsers - even those not
supporting 'text-decoration-line: overline' - would be said to pass such
test.
'border-top: solid 1px black' is not stated or said anywhere that it is
an equivalent to 'text-decoration-line: overline'.

Proposal:

<link rel="help" title="CSS Text Decoration Module Level 3: 2.1. Text
Decoration Lines: the 'text-decoration-line' property"
href="http://www.w3.org/TR/2012/WD-css-text-decor-3-20121113/#text-decoration-line-property"
/>
(...)
		<meta name="flags" content="" />
(...)
		<style type="text/css"><![CDATA[
				div
				{
					text-decoration-line: overline;
				}
		]]></style>
	</head>
	<body>
		<p>Test passes if "Text sample" has an horizontal line over it.</p>
		<div>Text sample</div>

and not to do a reftest for this test.

6-
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/text-emphasis-position-above-left-001.xht
and other text-emphasis-position-*.xht tests

"
Test passes if a pair of upper and lower block in the square box is
identical.
"

could be better replaced with

"Test passes if there are <strong>2 identical</strong> blue-and-yellow
rectangles."

Ideally, you want all pass/fail test conditions sentence to be easy to
understand, especially for non-CSS savvy people (your mother, my
neighbour, your bus driver, etc).

<p>horizontal</p>
should be removed or be made as a comment or qualifier in the text
assert and title.

7-
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/text-emphasis-style-shape-001.xht

http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/word-break-keep-all-001.xht

Generally speaking, long tests with multiple sub-tests are not ideal for
the Test Harness.

"
Short
    Tests should be very short and certainly not require scrolling on
even the most modest of screens
"
http://wiki.csswg.org/test/format#design-requirements

8-
http://test.csswg.org/source/contributors/east-tokyo/submitted/css3-text/text-indent-percentage-001.xht

"
If the test uses the Ahem font, make sure its computed font-size is a
multiple of 5px
"
http://wiki.csswg.org/test/format#acceptable-test-formats

"Test passes if the following two text blocks look same in terms of
margin-left and text-indent respectively."

could be better replaced with

"Test passes if the following 2 paragraphs have the exact same
horizontal position."

And then use 2 real text (eg "Lorem ipsum ...") paragraphs instead of
letters from the alphabet.

Finally, theoretically speaking, it is possible for a property to round
down a fraction of a pixel while another my round up a fraction of a
pixel. So, it's not entirely sure, certain that margin-left: 50% can be
equivalent to, can use the same horizontal space taken with text-indent:
50%.



Gérard
-- 
Contributions to the CSS 2.1 test suite:
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/

CSS 2.1 Test suite RC6, March 23rd 2011:
http://test.csswg.org/suites/css2.1/20110323/html4/toc.html

CSS 2.1 test suite harness:
http://test.csswg.org/harness/

Contributing to to CSS 2.1 test suite:
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/web-authors-contributions-css21-testsuite.html

Received on Monday, 21 January 2013 22:26:11 UTC