[webkit-dev] DOM L2 HTML self-hosted tests
Curt Arnold
curt.arnold at mac.com
Mon Jun 27 07:49:38 PDT 2005
From: curt.arnold at mac.com
Subject: Re: [webkit-dev] DOM L2 HTML self-hosted tests
Date: June 27, 2005 2:32:22 AM CDT
To: curt.arnold at mac.com
On Jun 25, 2005, at 8:03 PM, Curt Arnold wrote:
> Apparently, self-hosted tests for L2 HTML were not previously
> generated, or if they had been the build had been broken for a
> while. I've worked a couple of days getting the build process
> working for them and will commit the changes to the W3C CVS likely
> tomorrow.
>
>
Committed changes to the W3C CVS to fix the self-hosted productions
for L2 HTML test suite. Updated my "unofficial nightly" build at
http://homepage.mac.com/curt.arnold/FileSharing1.html. Should be
simple (but time-consuming) to build from the CVS. Download Ant
1.6.5 from http://ant.apache.org then:
cvs -d :pserver:anonymous at dev.w3.org:/sources/public login
anonymous
cvs :pserver:anonymous at dev.w3.org:/sources/public co 2001/DOM-Test-Suite
setenv ANT_OPTS -Xmx900M
cd 2001/DOM-Test-Suite
ant dom2-html-gen-selfhtml
ANT_OPTS is needed to increase memory in Java VM to avoid "Out of
Memory" error during transforms.
> HTMLDocument04-expected.txt: text run at (2,2) width
> 40: "failure"
>
> Failed consistently by Firefox and Safari in self-hosted, but
> passed in JSUnit.
> Looks like self-hosted rendering error. Will review.
>
>
...
> HTMLScriptElement01-expected.txt: text run at (2,2)
> width 40: "failure"
> HTMLScriptElement02-expected.txt: text run at (2,2)
> width 40: "failure"
> HTMLScriptElement03-expected.txt: text run at (2,2)
> width 40: "failure"
> HTMLScriptElement04-expected.txt: text run at (2,2)
> width 40: "failure"
> HTMLScriptElement05-expected.txt: text run at (2,2)
> width 40: "failure"
> HTMLScriptElement06-expected.txt: text run at (2,2)
> width 40: "failure"
> HTMLScriptElement07-expected.txt: text run at (2,2)
> width 40: "failure"
>
> All appear to be production problems with self-hosted tests.
> Firefox passes them
> in JSUnit, fails them self-hosted. Understandable since "self-
> hosting" injects new
> script elements into test documents which would likely disrupt
> finding the script
> element under test.
>
The build.xml file will correct for the changed expectations due to
the change of the document URL and injection of additional script
elements. After those changes, Safari still fails
"HTMLScriptElement01". The test expects HTMLScriptElement.text to
contain the text of the script which Firefox, IE and Opera return.
Safari returns "".
More information about the webkit-dev
mailing list