Please see attached reduction. It has two instances of the anchor tag, both with a background image set via CSS. The first has no text inside the tag, the other does. Both render in Firefox, WebKit does not render the first since is has no text, ignoring the background image CSS.
Created attachment 13896 [details] anchor tag reduction
We skip empty inlines that are within ignored whitespace. This is being debated in CSS2.1 right now but the spec will probably say we should render it.
May be related to: http://bugs.webkit.org/show_bug.cgi?id=13995
Created attachment 15654 [details] This file contains the patch for bug-id 13237. There 5 layout tests failing with this fix. But, these are valid failure cases considering the bug in hand. The following 5 test cases are failing in the layout test. (1) fast/dom/onerror-img.html (2) fast/forms/access-key.html (3) fast/lists/markers-in-selection.html (4) tables/mozilla/other/wa_table_thtd_rowspan.html (5) tables/mozilla/other/wa_table_tr_align.html
If the changes in layout tests are correct, then the patch should include updated test results.
(In reply to comment #5) > If the changes in layout tests are correct, then the patch should include > updated test results. > Darin, I did not understand what you have stated above. I have not modified any existing Layout test. I have added one layout test specific to this bug. My patch includes this test case and the expected.txt for the same. How do I include the updated test results for the 5 test cases, discussed in Comment#4, in my patch ?
(In reply to comment #6) > (In reply to comment #5) > > If the changes in layout tests are correct, then the patch should include > > updated test results. > > Darin, I did not understand what you have stated above. > I have not modified any existing Layout test. > I have added one layout test specific to this bug. > My patch includes this test case and the expected.txt for the same. > > How do I include the updated test results for the 5 test cases, discussed in > Comment#4, in my patch ? If the test results are different, you may run the WebKit tests using: WebKitTools/Scripts/run-webkit-tests --reset-results
Comment on attachment 15654 [details] This file contains the patch for bug-id 13237. This is completely wrong.
There is absolutely no reason to add special-case "anchor element" code to bidi.cpp.
offsetLeft/Top are relative to your offsetParent. I would guess we had offsetParent wrong. The goal here is to match IE. What does IE return here?
(In reply to comment #10) > offsetLeft/Top are relative to your offsetParent. I would guess we had > offsetParent wrong. The goal here is to match IE. What does IE return here? I believe that comment was meant for Bug 14722.
(In reply to comment #10) > offsetLeft/Top are relative to your offsetParent. I would guess we had > offsetParent wrong. The goal here is to match IE. What does IE return here? > I don't understand how offsetLeft/Top/Parent affect this bug. Could you please explain more on it ? IE doesn't render properly to this scenario. It leaves an empty space(without a boundary) for the empty anchor element, where as it renders a box to the anchor tag having some text inside. Firefox renders a box of minimal width for an empty anchor tag and renders a box of required width if an anchor element has some content within.
(In reply to comment #10) > offsetLeft/Top are relative to your offsetParent. I would guess we had > offsetParent wrong. The goal here is to match IE. What does IE return here? > Dave, was your comment intended for bug #13237 or for bug #14722 ? David Kilzer thinks you may have accidentally posted comment #10 to bug #13237. If your comment was intended for bug #14722, should the behavior for bug #13237 match IE or Firefox? (for SAP we need the behavior to match Firefox)
<rdar://problem/5399614>
*** Bug 13995 has been marked as a duplicate of this bug. ***
Fixed with r27254.
Created attachment 19263 [details] test firefox
Created attachment 19264 [details] test csv safari