Re: [css3-ui] cursor image format tests

Hello Gérard,

Monday, March 23, 2015, 8:43:23 PM, you wrote:

> Le 2015-03-23 12:01, Chris Lilley a écrit :
>> Hello Public-css-testsuite,
>> 
>> Spurred on by recent discussions around mandated image formats for the
>> CSS cursor property, I have submitted some tests.
>> 
>> http://test.csswg.org/shepherd/search/testcase/spec/css-ui-3/section/6.1.1/load/t53/#t16
>> 
>> A) cursor-image-001.html to cursor-image-009.html are basic tests just 
>> to
>> have a quick look at image format support for cursors:
>> 
>> 001   PNG image with css-supplied hotspot, relative URL.
>> 002   PNG image with css-supplied hotspot, absolute URL.
>> 003   ICO cursor with its own hotspot, relative URL.
>> 004   CUR cursor with its own hotspot, relative URL.
>> 005   SVG cursor with CSS-supplied hotspot, relative URL.
>> 006   ANI cursor with its own hotspot, relative URL.
>> 007   Compressed (SVGZ) cursor with CSS-supplied hotspot, relative URL.
>> 008   Non-existent image with CSS-supplied hotspot, relative URL, and
>> help cursor fallback
>> 009   PNG image with css-supplied hotspot, relative URL, no fallback.
>> 

> Chris,

> 1-
> We want tests to be using simple language, simple vocabulary so that 
> ordinary people (your mother, my neighbour, children, grandparents who
> never had a personal computer) could take the tests and enter results.

Balanced against that, we also want the reported results to be
accurate and for it to be crystal clear exactly what the test is
testing.

> So, words/expressions like "cursor" (some people think that a cursor is
> the blinking white underscore-looking thing in black background text 
> environment ... which is also true!),

Adding a tutorial on what a cursor is seems outside the scope of a
test suite. I'm testing cursors. If the person running the test does
not know what a cursor is, we don't want their results.

>  PNG, "css-supplied hotspot",
> "absolute URL", etc... should not be seen by people taking tests. But 
> they should be in the text assert and/or in comments in the test or in
> <title> text.

I'm fine to put that detail into the assertions.

I'm not fine to allow test submissions by people who don't know what a
computer is, or what a rectangle is, etc.

> 2-
> pale green: there is no reason to have a background-color here

Its spring grass for the sheep. More seriously, a cursor which is
mostly white with some black and pink is much more visible against a
coloured background.

You will also see that mostly I used a mid grey, but changed that 9to
peach) for tests with grey or grey+alpha cursors so that it was easier
to see the correct result.

> 3-
> width set to 8em: why not just let it be 'auto'?

The intent was to ensure that the largest cursor plus largest
reference image would fit into the box, while also making sure the box
is fairly small; the cursor can then be compared more easily inside
and outside the target area. I agree though that the value should be
changed; as we don't set a font size and the cursors are sized in
pixels, the width and height of the test area should be in px.

> 4-
> why not just a black border instead of #555?
I don't see how a black border makes the test any better.

> Here's a test that couldn't make it before RC6 in 2011 and that I forgot
> to submit:

> http://www.gtalbot.org/BrowserBugsSection/css21testsuite/cursor-025.html

> You could create different "PASS" cursors (same look, same colors) in 
> different formats; that way, people would even be able to anticipate 
> expected results in a serie of tests.

Nice PASS cursor. Your test area is too large, though. Getting back to
your initial points, it also requires people to understand "hover" and
"pointing device". And the pointing device (trackpad, mouse, etc) does
not change - the cursor changes. So I think my wording on the test
pass criteria is better than yours, there.

> The kind of tests - ideally - you want to create as a test author are 
> tests where people do not have to think.

> Other points:

> 5- Absolute URL (outside, different domain name test.csswg.org): you 
> will need to get Peter Linss approval on this... as I am not sure this
> is okay

Sure, happy to do that.

> 6- The fallback cursor should be 'auto' and not 'help' in those cursor
> tests.
Because ....

Help is better because it is not the default value. So it is easier to
see if the fallback has been used, rather than some auto value. Making
failure more evident makes for a better test.

> 7- I think you should be testing sweet spot or hot spot in distinct 
> tests, in separate tests from various image format.
Here I am just testing that the syntax doesn't cause the cursor to
fail. I agree that separate tests are needed that actually check the
hotspot values are being used.

> 8- Like Florian wrote, BOM-UTF-8 should be added, image flag should be
> added
Right.

> 9- Just a <div> instead of <p class="test">. Class is for logical 
> grouping of various elements; id is for a document-unique element. We 
> suggest to use <div id="test"> when the test includes a correspondent,
> associated reference (div id="reference"). Here, your tests do not have
> such correspondent, associated reference. So, there is no need to add an
> id attribute. Just a <div></div> is good enough.
OK.

> 10- 006   ANI cursor: we already have cursor .ani test in CSS 2.1 test
> suite.
> http://test.csswg.org/suites/css2.1/nightly-unstable/html4/cursor-024.htm

Since that test does not have clear test assertions or spec links,
mine is better for css3-ui..



-- 
Best regards,
 Chris Lilley, Technical Director, W3C Interaction Domain

Received on Tuesday, 24 March 2015 12:05:07 UTC