Hover over the image in the testcase. You will see that the second image loads again, although it had already been loaded for cross-fade() to work. This is wasteful and should be rectified.
<rdar://problem/10946032>
Created attachment 137083 [details] reduced test case reduced test case by removing cross-fade (which appears to no longer work in firefox, btw). We load 2 images, and when mouse hover on the first image, we change it to the second image. In firefox, this change is instantaneous, but in webkit, this change reloads the image. the issue has nothing to do with cross-fade itself, but has to do with the fact that cached resources is not reused when there are no cache headers present. Specifically, CachedResource::freshnessLifetime currently returns 0. The comment says that other browsers do this, but it seems like firefox doesn't. I plan to fix this by bumping it up from 0 to 3600 (i.e. 1 hour). I'm open to alternate suggestions though.
Please re-title this bug to reflect the actual issue.
Created attachment 137100 [details] Patch returning 3600 in freshnessLifetime if no cache headers present
Antti should review this.
What is the current behavior in other engines? It seems bit risky to change this if everyone else considers these resources uncacheable.
Ah you mentioned Firefox. Wonder what their exact behavior is?
I'm not sure what their exact behavior is, i'll investigate this further
Comment on attachment 137100 [details] Patch Attachment 137100 [details] did not pass chromium-ews (chromium-xvfb): Output: http://queues.webkit.org/results/12405044 New failing tests: http/tests/misc/script-async.html http/tests/xmlhttprequest/access-control-basic-allow-preflight-cache.html
Created attachment 137123 [details] Archive of layout-test-results from ec2-cr-linux-01 The attached test failures were seen while running run-webkit-tests on the chromium-ews. Bot: ec2-cr-linux-01 Port: <class 'webkitpy.common.config.ports.ChromiumXVFBPort'> Platform: Linux-2.6.35-28-virtual-x86_64-with-Ubuntu-10.10-maverick
This is definitely not a CSS bug.
Comment on attachment 137100 [details] Patch This change needs supportive evidence. Please feel free to ask for review again with evidence provided.