WK2 EFL WebKitTestRunner currently has a very basic implementation of TestRunner::pathToLocalResource() method (it merely returns the argument). We need to map /tmp to ${DUMPRENDERTREE_TEMP} envar and /tmp/LayoutTests to ${LOCAL_RESOURCE_ROOT} envar as we do for WK1 EFL DumpRenderTree. This causes some test cases for fail for WK2-EFL, including: fast/dom/frame-loading-via-document-write.html
Created attachment 162228 [details] Patch
Comment on attachment 162228 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=162228&action=review > Tools/WebKitTestRunner/InjectedBundle/efl/TestRunnerEfl.cpp:62 > JSRetainPtr<JSStringRef> TestRunner::pathToLocalResource(JSStringRef url) > { > - return url; > + String requestedUrl(url->characters(), url->length()); Shouldn't you verify it is local? assert? > Tools/WebKitTestRunner/InjectedBundle/efl/TestRunnerEfl.cpp:66 > + if (requestedUrl.find("LayoutTests") != notFound) { Shouldn't it start with it? or is that not needed
Comment on attachment 162228 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=162228&action=review >> Tools/WebKitTestRunner/InjectedBundle/efl/TestRunnerEfl.cpp:62 >> + String requestedUrl(url->characters(), url->length()); > > Shouldn't you verify it is local? assert? The function is called pathTo*Local*Resource() so I assumed it was not needed. FYI, I merely copied the implementation from DumpRenderTree. >> Tools/WebKitTestRunner/InjectedBundle/efl/TestRunnerEfl.cpp:66 >> + if (requestedUrl.find("LayoutTests") != notFound) { > > Shouldn't it start with it? or is that not needed Technically, it needs to start with file:///tmp/LayoutTests. This is a copy of the DRT implementation but I can improve it.
Comment on attachment 162228 [details] Patch Clearing flags on attachment: 162228 Committed r127602: <http://trac.webkit.org/changeset/127602>
All reviewed patches have been landed. Closing bug.