RE: Capitalization issues

I have to say I haven't really been following this discussion much and 
Rick is on vacation, but I did poke around a bit and found the following
message, which gives some ideas on the original intent of these tests.
I'm not sure how they have been modified since the initial release, but 
hopefully this will help...I'll try to look a little closer as well.

http://lists.w3.org/Archives/Public/www-dom-ts/2002Mar/0020.html

--Mary


-----Original Message-----
From: www-dom-ts-request@w3.org [mailto:www-dom-ts-request@w3.org] On
Behalf Of Curt Arnold
Sent: Wednesday, June 25, 2003 1:07 PM
To: David Faure
Cc: www-dom-ts@w3.org; staikos@kde.org
Subject: Re: Capitalization issues


David Faure wrote:

>
>Hello,
>
>Many tests fail in Konqueror (and almost certainly Safari, maybe other 
>browsers too) due to a capitalization issue.
>For instance hc_attrname.html expects "class" but since the document
>is HTML, we return "CLASS", as per the HTML specification.
>
>According to Curt Arnold this would be (from memory) because the
>spec says uppercase, but since Mozilla and IE return lowercase,
>the tests were changed to expect lowercase.
>
>But this completely screws up any browser that attempts to actually
comply
>to the DOM spec, which is what this is all about, isn't it?
>If the tests accept that Mozilla and IE return lowercase, why not
simply
>accept both (the "Mozilla and IE" way, and the "compliant" way, i.e.
uppercase)?
>
>
>PS: congratulations for the improvements to the test suite. It looks
much
>more complete now, it catches many little unimplemented things - which
>I'm fixing one by one :)
>
>  
>
I'm digging into this right now, my recollection is fuzzy and I've not 
been able to find the "spec doesn't say what it appears to say" message 
that I recalled.

David's observation is about test failures with Level 1 Core hc_* tests.


hc_attrname is an example of one of the tests that is failing on 
Konqueror where it checks the nodeName of a "class" attribute.  The 
original version of hc_attrname.xml 
(http://dev.w3.org/cvsweb/2001/DOM-Test-Suite/tests/level1/core/hc_attrn
ame.xml) 
which was checked in by dom_ts_2 (Rick Rivello?) on 8 March 2002 
specified ignoreCase="auto" which would have required an HTML 
implementation to return "CLASS" (as Konqueror is now doing).   On 10 
June 2002 (as part of a massive update to the HTML compatible Core 
tests), I checked in a modified version that specified 
ignoreCase="false" that would require an HTML implementation to return 
"class" (which is the case of the attribute in the source document). 

By my reading of 
http://www.w3.org/TR/DOM-Level-2-HTML/html.html#ID-535378264, 
ignoreCase="auto" (upper case for HTML implementations, lower case for 
XHTML) would be correct.

I have recollections of ignoreCase being misused 
(http://lists.w3.org/Archives/Public/www-dom-ts/2002Mar/0020.html point 
5) in the initial L1 Core HTML compatible tests.  So the change from 
ignoreCase="auto" to ignoreCase="false" might be a side-effect of a 
global change to reset everything to the most restrictive setting with 
the intention of later restoring those that should have had other
settings.

Here is a listing of previous messages that touched on the subject of 
case in HTML DOM's:

http://lists.w3.org/Archives/Public/www-dom-ts/2002Mar/0056.html
http://lists.w3.org/Archives/Public/www-dom-ts/2002Jun/0039.html
http://lists.w3.org/Archives/Public/www-dom-ts/2002Feb/0009.html
http://lists.w3.org/Archives/Public/www-dom-ts/2002Feb/0023.html
http://lists.w3.org/Archives/Public/www-dom/2002JanMar/0093.html
http://lists.w3.org/Archives/Public/www-dom/2002JanMar/0095.html
http://lists.w3.org/Archives/Public/www-dom/2002AprJun/0171.html
http://lists.w3.org/Archives/Public/www-dom-ts/2003Feb/0012.html
http://lists.w3.org/Archives/Public/www-dom-ts/2003Feb/0020.html
http://lists.w3.org/Archives/Public/www-dom/2001JulSep/0179.html


I'm continuing to explore this issue.

Received on Wednesday, 25 June 2003 14:05:12 UTC