RE: [CSS21] Test Suite

Ian (and Tantek),

First, thanks for publishing the link to the early CSS21 acceptance tests
suite.  I've gone through them, and I have a few comments I hope will be
useful.

Comments:
=========

1)  The '.xml' extension doesn't match any of the proposed extensions
documented in Section 3.7 of the CSS2.1 Test Case Authoring Guidelines. [1]

Unless there is some other compelling reason, I would suggest changing the
extension for XHTML tests to 'xhtml'.

Due to various limitations (primarily of IE6), neither the 'xhm', nor the
'xml' extensions are particularly good.  If the tests are renamed with the
xhtml extension, the tests can be loaded using Firefox, Mozilla, Opera,
*and* IE6.

For the interested, '.xml' files cause IE6 to parse the identified DTD
(xhtml11.dtd).  The DTD for XHTML 1.1 is modular, and IE6 incorrectly barfs
on the missing but IGNORE'd xhtml-prefw-redecl.mod.

In addition, IE6 uses the "file extension" to determine which application
should handle the received content even though URI's don't have file
extensions.  For URI's ending in '.xht', IE6 attempts to open the (unknown)
associated application, and the user is presented with a useless "Open
With..." dialog.


2)  The tests correctly enclose the content of the <style> element in
'<![CDATA[' and ']]>' CDATA section markers [2], but few of the current
browsers handle this syntax (probably due to the incompatible definition of
<style> element content between HTML 4, and XHTML 1).  The effect is that
the first rule is ignored (or all rules in the case of Firefox).

I would suggest eliminating the CDATA section markers from the tests since
this exercises a feature of the XHTML parser, and shouldn't affect the
results unless a '<', or '&' are included within the content of the <style>
element.  Although css1test11.xml contains a '<', it works without the CDATA
section markers because the XHTML parser sees this as a comment, but the CSS
parser sees it as a CDO token.


3) PNG images are used in 75 of the tests, while GIF images are used in 4.
The use of PNG's doesn't add anything to the tests, yet eliminates UA's that
don't render this format (most notably IE6).

I would recommend converting 28 distinct PNG's to GIF.  I would be happy to
provide this conversion if that would help.

If you elect *not* to convert the PNG's, I would suggest converting the
remaining GIF's so you can minimize the requirements of the UA's rendering
engine.


4) A number of the tests use the 'Ahem' font (93 to be precise).  Although I
can understand the convenience of the technique, it relies heavily on the
particular handling of font rendering which is explicitly left to the
implementer in places (e.g. height of inline, non-replaced elements [3]).
Worse, it implicitly requires all conformant UA's to implement a font
mechanism that can accept the Ahem font, port the Ahem font to their
platform, or rewrite the 93 tests according to the "interoperable'
definition. [4]

I would suggest carefully limiting the number of tests that rely on the Ahem
font.  I believe most (if not all) of the Ahem tests can be rewritten to
accomplish the same effects using inline replaced elements.  Clearly,
precise testing of the font-size property will not benefit from inline
images, but there should be ways to test this property without requiring
Ahem.

I would also suggest that a link to the Ahem download page [5] be included
either in the body, or in the head of any test that does require the Ahem
font.


[1] http://www.w3.org/Style/CSS/Test/guidelines.html#filenames
[2] http://www.w3.org/TR/2002/REC-xhtml1-20020801/#h-4.8
[3]
http://www.w3.org/TR/2004/CR-CSS21-20040225/visudet.html#inline-non-replaced
[4] http://www.w3.org/TR/2004/CR-CSS21-20040225/cover.html
[5] http://www.hixie.ch/tests/evil/mixed/lineheight4.html



> -----Original Message-----
> From: Ian Hickson [mailto:ian@hixie.ch]
> Sent: Thursday, July 08, 2004 8:03 AM
> To: Mark Moore
> Cc: tantekc@microsoft.com
> Subject: Re: [CSS21] Test Suite
> 
> On Fri, 2 Jul 2004, Mark Moore wrote:
> >
> > According to the CSS2.1 CR, a test suite is required as one of the key
> > exit criteria, but I haven't been able to find either the suite, or the
> > current status of features, implementations, or interoperability.
> 
> The suite is being written. I have a work-in-progress dump at:
> 
>    http://www.hixie.ch/tests/evil/css/css21/source/raw-tests/
> 
> ...but it is far from complete.
> 
> The current status of features, implementations, and interoperability is
> changing on a daily basis and is not tracked anywhere to my knowledge.
> 
> --
> Ian Hickson               U+1047E                )\._.,--....,'``.    fL
> http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
> Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Wednesday, 21 July 2004 18:25:36 UTC