[webkit-dev] Questions about expected test results

[Sami Kukkonen]

I'm interested in exploring what it would take to produce a new linux
port to something other than Qt or GTK. As a starting point I wanted to
check out a reasonably stable revision to start my exploration.

 

Buildbot trunk-qt-linux-release build #13700 was all green so I checked
out revision 41211 and performed a linux Qt 4.4.0 build on Ubuntu 8.04.
Build succeeded without problems but when I executed run-webkit-tests I
was surprised to discover that 445 test cases failed and 13 crashed.
Then I noticed that buildbot only executes jscore-test and no layout
tests. So I executed run-javascriptcore-tests and got only 1 failure so
that's much better. On the other hand, when I looked at results file it
contained 51 failures.

 

Being new to WebKit this is confusing so I have a couple of questions:

 

1.	Why doesn't trunk-qt-linux-release run layout tests? Is it known
and expected that hundreds of them will fail? 
2.	Why does trunk-qt-linux-release report jscore-test as passed
when results file contains 50 failures?

 

/Sami

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20090225/c9883805/attachment.html>