Though LINK_PREFETCH feature is enabled, fast/dom/HTMLLinkElement/subresource.html is still failed because of below difference, - prefetch.link has MIME type application/octet-stream + prefetch.link has MIME type text/plain
In WebViewHost class of chromium port, didReceiveResponse() returns "application/octet-stream" when MIME type is empty as below. void WebViewHost::didReceiveResponse(WebFrame*, unsigned identifier, const WebURLResponse& response) { ... if (m_shell->shouldDumpResourceResponseMIMETypes()) { GURL url = response.url(); WebString mimeType = response.mimeType(); printf("%s has MIME type %s\n", url.ExtractFileName().c_str(), // Simulate NSURLResponse's mapping of empty/unknown MIME types to application/octet-stream mimeType.isEmpty() ? "application/octet-stream" : mimeType.utf8().data()); } } However, in EFL port case, we just print real mime_type, which has text/plain. void DumpRenderTreeChrome::onResponseReceived(void*, Evas_Object*, void* eventInfo) { ... if (!done && gLayoutTestController->dumpResourceResponseMIMETypes()) { printf("%s has MIME type %s\n", KURL(ParsedURLString, response->url).lastPathComponent().utf8().data(), response->mime_type); } } It looks there is some implementation difference between chromium port and EFL port. Thus, it seems EFL port needs to have new expected result.
Created attachment 156964 [details] Patch
CC'ing Hajime, could you take a look this patch? I'm not sure if EFL port can have new expected result for subresource.html. If default mime type can be changed when mime type is empty in chromium port, I think EFL port doesn't need to have new expected result for subresource.html.
Comment on attachment 156964 [details] Patch Clearing flags on attachment: 156964 Committed r124946: <http://trac.webkit.org/changeset/124946>
All reviewed patches have been landed. Closing bug.