W3C home > Mailing lists > Public > public-css-testsuite@w3.org > June 2007

Re: Assumptions about UA default style sheets

From: fantasai <fantasai.lists@inkedblade.net>
Date: Thu, 28 Jun 2007 01:46:28 -0400
Message-ID: <46834B34.1000706@inkedblade.net>
To: "Grant, Melinda" <melinda.grant@hp.com>
CC: public-css-testsuite@w3.org

Grant, Melinda wrote:
> I wonder if we should import a style sheet specifying the
> recommendations in Appendix D in all tests as a starting point, to avoid
> differences in results based on different UA style sheet defaults.
> 
> I am currently responding to a complaint from another company regarding
> t1402-c45-bg-canvas-00-b.xht and t09-c5526c-display-00-e.xht. This
> company uses a display value of 'inline-block' for the 'html' element in
> their default style sheet, which apparently causes them to fail these
> tests. (Language barriers make it difficult to understand why, but
> there's nothing normative prohibiting it, sfaik.)

There is nothing in the spec that would allow them to fail
   t1402-c45-bg-canvas-00-b.xht
due to "html {display: inline-block}". If they've got a bug that's exposed
because they do that, then we should certainly not fix the test case so
that they can pass it.

> We could always modify the tests to add
> 	html {display: block}...
> ...but what else might we need to add to allow all compliant
> implementations to pass?

I'd rather create a new test that adds html { display: inline-block; }
to see if any other implementations fail. :)

> Importing the style sheet in App D would remove these variances.
> 
> I know this has the downside of burdening all tests with an imported
> style sheet; but I think this is more of a theoretical issue than a
> practical one; whereas differences in default style sheets *should* be
> allowed without causing tests to fail.
> 
> Including the rules in App D directly within the test file would, I
> think, create too much visual "noise". So importing a style sheet seems
> the least evil to me.
> 
> Thoughts / other suggestions?

I would rather continue with the WG's prior decision on this issue, which
was to deal with it on a case-by-case basis. Not only would importing the
entire Appendix D stylesheet complicate our tests unnecessarily, it won't
solve the problem where a UA fails because it adds *extra* rules to its
UA style sheet. Appendix D does not reset everything to its initial value.

~fantasai
Received on Thursday, 28 June 2007 05:46:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 September 2010 17:51:55 GMT