Should the RenderInline expand to enclose the float? It currently gets a zero height - which is why the RenderBlock parent gets a zero height.
Just a guess but: static bool alwaysRequiresLineBox(RenderInline* flow) { // FIXME: Right now, we only allow line boxes for inlines that are truly empty. // We need to fix this, though, because at the very least, inlines containing only // ignorable whitespace should should also have line boxes. return !flow->firstChild() && flow->hasInlineDirectionBordersPaddingOrMargin(); } I think !flow->firstChild() (the subject of the FIXME) is probably why we're failing this test. You should replace that with a function that recurs and looks for whether or not you have child content that will force the creation of a line box.
This test can be at least partly reduced to the fact that the inline div will get a line box in this case: <div><div style="display:inline"></div></div> But not in this: <div><div style="display:inline"><span></span></div></div> FF, Opera and IE create a line box for the latter case. So presumably WebKit should too. But then WebKit would always create a line box for empty lines. Is that what should happen?
Created attachment 184913 [details] Patch
Comment on attachment 184913 [details] Patch Attachment 184913 [details] did not pass win-ews (win): Output: http://queues.webkit.org/results/16155267
Comment on attachment 184913 [details] Patch Attachment 184913 [details] did not pass mac-ews (mac): Output: http://queues.webkit.org/results/16155308 New failing tests: fast/text/international/bidi-ignored-for-first-child-inline.html
Comment on attachment 184913 [details] Patch Attachment 184913 [details] did not pass mac-wk2-ews (mac-wk2): Output: http://queues.webkit.org/results/16146538 New failing tests: http/tests/inspector/resource-tree/resource-request-content-while-loading.html fast/text/international/bidi-ignored-for-first-child-inline.html
Created attachment 185011 [details] Patch
Comment on attachment 185011 [details] Patch Attachment 185011 [details] did not pass mac-ews (mac): Output: http://queues.webkit.org/results/16180278 New failing tests: fast/text/international/bidi-ignored-for-first-child-inline.html
Comment on attachment 185011 [details] Patch Attachment 185011 [details] did not pass mac-wk2-ews (mac-wk2): Output: http://queues.webkit.org/results/16191013 New failing tests: fast/text/international/bidi-ignored-for-first-child-inline.html
Comment on attachment 185011 [details] Patch r=me
Committed r142152: <http://trac.webkit.org/changeset/142152>