I have tested Safari using the latest webkit (15th Feb) and it fails the following tests regarding paddings:
I think that the reason why you are seeing failures is that you don't have Ahem font installed (<http://webkit.opendarwin.org/quality/Ahem.ttf>).
This special simple font is used to construct certain types of predictable tests. A page on Ian Hickson's site <http://www.hixie.ch/resources/fonts/> has a more detailed explanation.
Indeed you were right. That has chanaged the appearances a lot. However there are still some small problems.
1px horizontal link on top of the green bar
lighter green line across the bar (exactly halfway)
Both show a small white column (maybe width=7px) between the colorful rows and the black right border
lighter green vertical line across the square (exactly halfway)
both show the lighter green vertical line across the bar (exactly halway)
there is also a red square which should be absent
I think the reporter would like this bug to be reopened per Comment #2.
Three of these tests are failing for me, and have identically broken results checked in:
Other tests start to break if I zoom in with Cmd+=. This might be an expected behavior.
Confirming this, but please note that CSS 2.1 is not fully supported and that this is a to-do.
Description: Add padding to CSS 2.1
This is still under discussion, however, i HAVE found this problem to be real myself as well.
All above mentioned tests renders correctly (except this test: http://www.w3.org/Style/CSS/Test/CSS2.1/current/t0804-c5509-ipadn-l-02-b-ag.htm) in Webkit r30001.
Other issues are related to antialiasing problem described in bug 17806. After turning font antialiasing off the issues disappear.
I mean bug 17086 :o
Created attachment 21016 [details]
I've created a patch for this, however the bidi.cpp is rather confusing, so I'm just hoping I did not break anything else... I ran all the tests from LayoutTests and nothing seems to be broken.
I'm happy to change whatever, since as I said, I barely knew what I was doing...
Created attachment 21018 [details]
A test case without padding
I don't understand:
1) Why the test is correct
2) How the failure is specific to padding (and therefore why the patch is correct)
Is WebKit's rendering of this test case correct? If not, I think that's what should be fixed, not anything specific to padding.
Created attachment 21032 [details]
test case for "spanned" string wrapping
I think the initial CSS 2.1 test exposes bug in Webkit which is completely unrelated to padding.
I my opinion Webkit incorrectly wraps "spanned" strings: "This_line<span>should_not</span>be_wrapped."
Comment on attachment 21016 [details]
This does not seem to address the root cause of the test failure.
Hmm, makes sense to me... For some reason I completely missed the fact that it was x<span>x</span> rathern than a padding issue...
Bug 18926 is essentially the same issue.