Created attachment 75576 [details] test case http://trac.webkit.org/changeset/72988 seems to have broken loading of iframes with pages that use @import somewhere in their style. Attached a test case steps to repro: open attached test case Expected behavior: the page with url http://www.nytimes.com/2010/11/07/education/07china-t.html?pagewanted=2&ref=education should load in the iframe and then shortly afterward the document is overtaken by http://www.nytimes.com/2010/11/07/education/07china-t.html. Not sure why, but anyway this is what happens in shipping safari. 5.0.3 TOT behavior: nothing loads in iframe. side note: in the test case is a commented-out line that uses http://www.rollingstone.com/politics/news/17390/119236?RS_show_page=1. This page doesn't use @import and behaves the same way in shipping as TOT. it also just loads in the iframe and doens't overtake the whole document in both cases. <rdar://problem/8728742>
See also bug 50532.
Pavel, had you had any chance to take a look? Is it reasonable to consider reverting r72988? The consequences of the bug fix seem more serious than the fixed symptoms.
*** Bug 50532 has been marked as a duplicate of this bug. ***
*** Bug 50526 has been marked as a duplicate of this bug. ***
*** Bug 50613 has been marked as a duplicate of this bug. ***
(In reply to comment #2) > Pavel, had you had any chance to take a look? Is it reasonable to consider reverting r72988? The consequences of the bug fix seem more serious than the fixed symptoms. Sorry, was flying across the ocean. Reverting this asap.
Committing to http://svn.webkit.org/repository/webkit/trunk ... M WebCore/ChangeLog M WebCore/css/CSSImportRule.cpp Committed r73427
Eventually we should land a regression test for this problem. Pavel, would you be willing to make one?
(In reply to comment #8) > Eventually we should land a regression test for this problem. Pavel, would you be willing to make one? Absolutely. We will need to fix the http://trac.webkit.org/changeset/72988 and it'll require a test.