Since the test passes when you run it on its own, I suspect the run-webkit-tests tool is to blame. run-webkit-tests fast/css/acid2.html --> pass run-webkit-tests fast/css/ --> fail remove everything in fast/css except for acid2, run-webkit-tests fast/css/ --> fail So maybe this has something to do with how the script handles directories? Here's the diff: --- /Volumes/Home/ggaren/Labyrinth/LayoutTests/fast/css/acid2-expected.txt 2005-10-17 17:16:47.000000000 -0700 +++ /tmp/layout-test-results/fast/css/acid2-actual.txt 2005-10-24 15:34:58.000000000 -0700 @@ -48,12 +48,11 @@ layer at (108,2686) size 96x12 RenderBlock (positioned) {BLOCKQUOTE} at (72,48) size 96x12 [border: none (24px solid #000000) none (24px solid #000000)] RenderBlock (floating) {ADDRESS} at (24,0) size 48x12 [bgcolor=#FFFF00] -layer at (84,2710) size 144x24 +layer at (84,2710) size 144x30 RenderBlock (positioned) {DIV} at (48,72) size 144x24 [bgcolor=#FF0000] RenderBlock {DIV} at (0,0) size 144x0 - RenderInline {OBJECT} at (0,0) size 131x14 - RenderInline {OBJECT} at (0,0) size 131x14 - RenderImage {OBJECT} at (13,0) size 131x24 [border: (12px solid #000000) none] + RenderInline {OBJECT} at (0,0) size 90x14 + RenderPartObject {OBJECT} at (54,0) size 90x30 RenderBlock (floating) {DIV} at (0,0) size 144x24 [border: (12px solid #FF0000) none (12px solid #000000)] RenderBlock {DIV} at (0,0) size 144x24 [border: (24px solid #FFFF00)] layer at (84,2782) size 144x24
This is super-irritating since it means we have a failing layout test. I don't think it's the run-webkit-tests script. I think instead it's something about reusing the same WebView used to load a previous test.
P1 because it's so inconvenient.
FWIW, a very unscientific test using nightly builds places the regression between 2005-10-20 19:27 PDT and 2005-10-21 19:27 PDT.
Geoff, can you still reproduce this problem?
Yes, I can still repro.
I should be more specific. I can still reproduce the underlying problem (fast/css and fast/css/acid2.html produce different results in the same manner they used to). However, it looks like someone checked in new test results so that run- webkit-tests now passes and run-webkit-tests fast/css/acid2.html now fails.
<rdar://problem/4404328>
Adding Regression keyword.
I think it's unlikely that this issue affects real-world websites, and it's not a practical day-to-day problem, so I'm removing the "P1" from this bug.
My fix for bug 7693 should finally fix this.