W3C home > Mailing lists > Public > public-css-testsuite@w3.org > January 2010

RE: Question on Microsoft's min-width-percentage-003.xht test

From: Gérard Talbot <css21testsuite@gtalbot.org>
Date: Wed, 6 Jan 2010 11:50:24 -0800
Message-ID: <595ff2a12e2be1ea77ea842cb090e42b.squirrel@cp3.shieldhost.com>
To: "Arron Eicholz" <Arron.Eicholz@microsoft.com>
Cc: "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>

> The point of the test case is to prove that the spec can actually be
> tested. All these tests are not necessarily written to test user agents
> they are first written to prove that a case can be written to actually
> see if it is feasible to get into the situation. This is such a case.
>
> The test is testing the specific assertion in the specification; though
> the pass condition text is incorrect. The pass conditions should state
> "Test passes if there is anything displayed below." I have just updated
> the test case with the updated pass conditions.


I understand Microsoft's approach to testing any statement or definition
or situation covered in CSS 2.1 spec which can be tested (or
testcase-able). Such approach is definitely systematic, methodical and
thorough and it is praiseworthy, impressive: I have said so publicly
twice in the past (and recently) to your colleague Dean Hachamovitch.
But here, I thought this systematic approach was going a bit too far.


> Note if we remove this case then we need a case that can prove that the
> text in the spec is valid. Or we need to remove the text since it isn't
> testable.

Realistically speaking, how can an user agent fail such test?

> If you can think of a better test than this great please
> submit it.

Personally, I would not have submitted such test, because, objectively
speaking, there is no fail condition.

"[ ] The test fails obviously whenever it fails."
CSS Test Review Checklist, Test design
http://wiki.csswg.org/test/css2.1/review-checklist

> At this point however this test covers the line in the spec.



Nota bene:

Mr Eicholz: One "nitpick" correction to do about

min-width-percentage-002.xht

http://test.csswg.org/source/contributors/microsoft/submitted/Chapter_10/min-width-percentage-002.xht

The meta assert says
"If the containing block's width is negative, ..."

but the containing block's width is:


            div
            {
                margin-right: -10px;
                width: 0;
            }

so negative or 0 has to be corrected.

best regards, Gérard Talbot
P.S.: I will be unable to reply to emails for the next few days.


> -----Original Message-----
> From: public-css-testsuite-request@w3.org
> [mailto:public-css-testsuite-request@w3.org] On Behalf Of Gérard Talbot
> Sent: Tuesday, January 05, 2010 4:17 PM
> To: public-css-testsuite@w3.org
> Subject: Question on Microsoft's min-width-percentage-003.xht test
>
> Hello,
>
> Regarding
>
> http://test.csswg.org/source/contributors/microsoft/submitted/Chapter_10/min-width-percentage-003.xht
>
> I would like to understand why such test. If the test assert says
>
> "
> If the containing block's width depends on this element's width, then
> the resulting layout is undefined.
> "
>
> (which is also what CSS 2.1 section 10.4
> http://www.w3.org/TR/CSS21/visudet.html#min-max-widths
> is saying)
>
> then what is such test testing exactly? What's the point (value,
> usefulness) of such test?
>
> If the words "Filler Text" are not there in some user agent, then how
> does such absence constitute a failure or some sort of spec violation?
>
> Am I missing something?
>
> regards, Gérard
Received on Wednesday, 6 January 2010 19:51:00 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 September 2010 17:52:01 GMT