Re: Latest DOM testsuite results.

David Faure wrote:

>Here is the summary of the outstanding issues with KHTML from KDE_3_1_BRANCH
>and the DOMTS from current CVS (thanks to Curt's provided tarball).
>
>15 failures out of 185 tests.
>
>To be done in KHTML:
>
>* Set m_specified in AttrImpl when the attribute is specified.
>See http://www.w3.org/TR/1998/REC-DOM-Level-1-19981001/level-one-core#ID-862529273
>(hc_attrspecifiedvalue.html, hc_attrspecifiedvaluechanged.html and hc_elementassociatedattribute.html)
>
>* Keep comments in the DOM tree (hc_commentgetcomment.html)
>
Mozilla and IE both fail that one.  Mozilla omits the comment node, IE 
reports the processing instruction and document type declaration as 
comments.  The <!DOCTYPE gets reported as if it were <!--CTYPE.  I can't 
see arguing that the IE behavior is conformant, but you could argue that 
dropping comments is an acceptible for HTML processors.

>
>* Debug some whitespace problems
>hc_elementnormalize.html and hc_textwithnomarkup.html [\n gets removed]
>hc_textparseintolistofelements: idem
>hc_textsplittextone.html and hc_textsplittextthree.html [Expected Jones but was ones]
>

IE fails this one in the same way.  These tests are based on the 
explicit XML whitespace processing rules and are passed by XML 
processors and Mozilla.  The current expected behavior must remain 
acceptible, however you could argue that IE and Konq current behavior is 
also acceptible since I don't think HTML specified as clearly how 
whitespace should be handled.

>
>===
>
>hc_nodeinsertbefore.html fails due to additional text node created by Mozilla (but not
>by IE nor Konqueror) with whitespace-only contents. DOMTS test bug -> filed as W3C BR #260.
>(affects also hc_nodelistindexgetlength.html and hc_nodelistreturnlastitem.html)
>  
>
Will investigate

>===
>
>Capitalization issues left (DOMTS bugs?)
>* hc_namednodemapreturnattrnode.html: class vs CLASS
>* hc_nodeattributenodename: title vs TITLE
>* hc_noderemovechildnode.html: em vs EM
>
>
>  
>
Most likely.  I'll fix these as a continuation of the existing bug.

Received on Monday, 30 June 2003 12:05:13 UTC