Re: [CSS21] Used value versus specified and computed value (position relative)

On 10/07/2012 16:57, "Gérard Talbot" wrote:
> Le Mar 10 juillet 2012 3:33, Anton Prowse a écrit :
>> On 09/07/2012 21:20, "Gérard Talbot" wrote:
>>> I'm struggling a bit with the understanding of following test:
>>>
>>> http://test.csswg.org/suites/css2.1/20110323/html4/relpos-calcs-001.htm
>>>
>>> What is inherited is the computed value, not the used value.
>>>
>>> So,
>>>
>>> line 25  top: inherit; /* using inheritance to test computed vs. used */
>>>
>>> top is 'auto'; top is not 50% and it is not 60px.
>>>
>>> So,
>>>
>>> 1- I do not see the purpose of the div.control in that test
>>>
>>> 2- div.inner's top property value inherited value is 'auto' which is
>>> resolved as 0; it is not 50%. div.inner's top property used value is 0,
>>> not 60px.
>>
>> If a UA incorrectly computed 'top' on div.inner to -50% (the used value
>> of that property on the parent) instead of 'auto' (the computed value of
>> that property on the parent) then that div would be positioned higher up
>> the canvas and the red background of div.control would show.
>
> Here's what I did. I set div.control's background color to orange (instead
> of red) and then set div.inner's top to -50%. I see red (from div.outer);
> I do not see orange (which would be from div.control).
>
> http://www.gtalbot.org/BrowserBugsSection/css21testsuite/relpos-calcs-001-Anton-P.html

Hmm, yes, the control div seems unnecessary.  One of those instances 
where the test author decided to test something else at the same time as 
the primary test?

>> A small nit about this test: div.container has been given a 'margin-top'
>> of '-60px', which is done purely for the aesthetic purpose of having the
>> square appear directly under the paragraph without a gap, but this
>> introduces the cognitive overhead of having to figure out whether or not
>> it's purely aesthetic in the first place.
>
> It's not purely aesthetic. [...] you want to create [...] a layout which can easily reuse
> the same reftest.

OK, mostly aesthetic ;-).  But I take your point about reusing the same 
reftest across multiple tests.

Cheers,
Anton Prowse
http://dev.moonhenge.net

Received on Tuesday, 10 July 2012 21:03:06 UTC