After http://trac.webkit.org/changeset/94593, fast/forms/textfield-overflow-by-value-update.html started failing: http://build.webkit.org/results/SnowLeopard%20Intel%20Debug%20(Tests)/r94596%20(2025)/fast/forms/textfield-overflow-by-value-update-pretty-diff.html http://build.webkit.org/builders/SnowLeopard%20Intel%20Debug%20(Tests)/builds/2025 http://build.webkit.org/builders/SnowLeopard%20Intel%20Debug%20(Tests)/builds/2023 I've manually bisected the builds to determine the revision.
The test depends on a JS file in the directory, but the contents of the failing test are: <!DOCTYPE html> <head> <script src="resources/common.js"></script> <style> input { -webkit-appearance: none; font-size: 32px; height: 0px; width: 100px; border-width: 0px; } </style> <body> <!-- The input value must not be shown outside the input border. So, the test result should be just white. --> <input id="textfield"> <script> $('textfield').offsetTop; $('textfield').value = "FAIL"; </script> </body>
Note this test is added by tkent in http://trac.webkit.org/changeset/90379.
The text result is not important for this test. If the pixel result is not changed, let's do rebaseline.
This actually doesn't fail for me in ToT Safari.
(In reply to comment #3) > The text result is not important for this test. If the pixel result is not changed, let's do rebaseline. (In reply to comment #4) > This actually doesn't fail for me in ToT Safari. Okay, that sounds great! I'm going to build ToT and run the pixel test to confirm that there's no regression.
I have confirmed that the pixel test still passes. I'm going to rebaseline it.
Created attachment 106532 [details] rebaseline the test
The test has failed on - "SnowLeopard Intel Debug (Tests)" bot - "Leopard Intel Debug (Tests)" bot The test hasn't failed on - "SnowLeopard Intel Release (tests)" bot - Chromium ToT bots on chromium.org - Chromium ToT bots on webkit.org - my local debug WebKit ToT on Snow Leopard
(In reply to comment #8) > The test has failed on > - "SnowLeopard Intel Debug (Tests)" bot > - "Leopard Intel Debug (Tests)" bot > > The test hasn't failed on > - "SnowLeopard Intel Release (tests)" bot > - Chromium ToT bots on chromium.org > - Chromium ToT bots on webkit.org > - my local debug WebKit ToT on Snow Leopard Mn... You're right. That is very weird.
It seems like the test is no longer failing.