W3C home > Mailing lists > Public > public-css-testsuite@w3.org > February 2012

Re: Requirements for (level >=3) tests

From: Aryeh Gregor <ayg@aryeh.name>
Date: Mon, 20 Feb 2012 12:31:59 -0500
Message-ID: <CAKA+Axkd7bze1xUWs0Hm=TDQtp6_Ck1iy874g_EwcfjFGJmUSQ@mail.gmail.com>
To: "Tab Atkins Jr." <jackalmage@gmail.com>, css21testsuite@gtalbot.org
Cc: "Linss, Peter" <peter.linss@hp.com>, Øyvind Stenhaug <oyvinds@opera.com>, CSS-testsuite <public-css-testsuite@w3.org>
On Fri, Feb 17, 2012 at 11:56 AM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
> Or alternately, *immediately* reserialize upon importing into the test
> system, so that diffs are against the relatively stable serialized
> formatting from the start.

Then it doesn't matter what format the tests are submitted in, minimal or not.

On Fri, Feb 17, 2012 at 3:12 PM, "Gérard Talbot"
<css21testsuite@gtalbot.org> wrote:
> "Inline styles should not be used unless the case is specifically
> testing this scenario."
> http://wiki.csswg.org/test/css2.1/format#style-element-embedded-styles

What's the reason for this requirement?  Using <style> for styles that
only affect one element in a small self-contained test case doesn't
make sense to me.  It means you have to look back and forth to find
what styles apply to what.  In fact, I saw one reftest (from Gecko
source code) that confused me considerably, because it had two divs
and the style wasn't inline, and the styles were in a different order
in the <style> than the divs were in the document, so I initially
thought the styles applied backwards.  So I actively prefer inline
style.
Received on Monday, 20 February 2012 17:32:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 February 2012 17:33:06 GMT