45586 2010-09-10 22:57:29 -0700 Having an empty listener to beforeload events changes the behavior of other scripts 2014-12-02 09:28:09 -0800 1 1 1 Unclassified WebKit DOM 528+ (Nightly build) Mac (Intel) OS X 10.6 RESOLVED FIXED http://odyniec.net/projects/imgzoom/ InRadar P2 Major --- 1 marc.hoyois aestes aa aestes ap brent.montrose fam.lam gavinp hyatt nicholas pkasting sam techsoftadvanced trev webkit oldest_to_newest 277650 0 67292 marc.hoyois 2010-09-10 22:57:29 -0700 Created attachment 67292 A trivial Safari extension reproducing the bug Overview: Simply registering a beforeload event listener in an injected script breaks the imgZoom jQuery plugin. Steps to reproduce: 1. Install the attached extension (and disable any other extension if you wish). It has the following trivial code in an injected script: document.addEventListener("beforeload", function(){}, true); 2. Navigate to http://odyniec.net/projects/imgzoom/ 3. Click on one of the three images. The image should enlarge. 4. Reload the page manually. 5. Click on the same image again. Actual result: The image shrinks. Expected result: The image should be enlarged as in step 3. Platform: Tested on OS X 10.6.4 with both Safari 5.0.2 and Webkit 6533.18.5, r67077 Note: Theoretically, this extension should have no effect whatsoever. The same bug happens with nontrivial event handlers. 278472 1 ap 2010-09-13 14:18:38 -0700 <rdar://problem/8424354> 282936 2 marc.hoyois 2010-09-22 08:07:04 -0700 The issue has been resolved by imgZoom 0.2.1. What was going on is that the script was using an image's offsetWidth property in a handler to the image's 'load' event. Somehow the presence of any beforeload event listener delays the setting of offsetWidth to just after the handler is called (as can be seen using a setTimeout of 0 milliseconds), but it is not set within the handler (i.e. equals 0). Since the spec does not require .offsetWidth to be set when the 'load' event is fired (the image's .width property has this purpose and works correctly), this behavior seems legit, although the fact that it happens exclusively with beforeload listeners indicates that it might be the consequence of another bug. 290033 3 aa 2010-10-05 18:50:35 -0700 (In reply to comment #2) > Since the spec does not require .offsetWidth to be set when the 'load' event is fired (the image's .width property has this purpose and works correctly), this behavior seems legit, although the fact that it happens exclusively with beforeload listeners indicates that it might be the consequence of another bug. I disagree that the behavior is legit. It appears that the behavior without anyone listening to the 'beforeLoad' event is that the width/height are always set in the 'load' event. I believe that is also the behavior in other browsers. Therefore the result of this bug is that extension developers can break web compat in a very unexpected way that would be difficult for them to anticipate, or to do anything about. 324000 4 trev 2010-12-17 14:55:38 -0800 The actual issue is that the "load" event for the image fires too early, before the image loads. In fact, the "load" event fires even before the "beforeload" event, you can clearly see it by putting console.log() calls into both event handlers. So you will always get a successful "load", even for an image that doesn't exists. And this is clearly not by the spec, I don't think there are any doubts about that. 324002 5 ap 2010-12-17 15:06:37 -0800 > the "load" event for the image fires too early, before the image loads. In fact, the "load" event fires even before the "beforeload" event Can you reproduce this with Web content (without an extension)? Could you please attach a test case? 324003 6 76914 trev 2010-12-17 15:06:55 -0800 Created attachment 76914 Minimal testcase Minimal testcase attached - you don't need an extension, a regular webpage will do. The first time you click the "load image" button (image not in cache yet) you get the correct output to the console: document's beforeload event, image width = 0 image's load event, image width = 100 Note that we *first* have "beforeload" and only then the "load" event. Second time you get: image's load event, image width = 0 document's beforeload event, image width = 0 Here and in all the subsequent runs you get the wrong behavior, "load" event fires before "beforeload" event. To get the original behavior you have to clear the browser cache, this bug apparently only happens for cached images. Tested with Chrome 8.0.552.224. 324007 7 ap 2010-12-17 15:18:18 -0800 Nasty. 454389 8 webkit 2011-08-21 04:17:44 -0700 This is still a significant bug that needs attention. See: http://code.google.com/p/google-ajax-apis/issues/detail?id=566 http://code.google.com/p/adblockforchrome/issues/detail?id=5821 457131 9 techsoftadvanced 2011-08-25 14:15:36 -0700 I am confirming that this is still a major bug, and much more so than the original report would have one believe. I am developing a rather large, multi-browser extension for an advertising company that blocks ads. (How that makes sense is a long story, and quite ingenious, but unnecessary for this discussion.) To summarize my findings, adding a trivial beforeload handler, at least as part of an extension, disables numerous scripts across the Web. Examples of notable sites that have become unusable after a trivial beforeload listener is installed include the Amazon AWS management console and Google Mail. This bug is currently known to inhibit the abilities of AdBlock Plus, a popular extension. I discovered this while looking to see how AdBlock Plus works around the bug. It doesn't. It can't. Rather than blocking resources from loading as it should and could if it weren't for this bug, it has to manually go through each element and remove any that attempts to load a blocked resource: a much less efficient method. As long as this bug persists, there is no way for extensions such as AdBlock Plus and my own to prevent resources from loading without predictably breaking an unacceptable number of web pages. 459634 10 trev 2011-08-31 07:59:54 -0700 Just realized that having the "beforeload" event handler around isn't necessary to get broken behavior. It is enough that the event handler was registered at some point - removing it again won't help you any more, the "load" event for images will still fire too early. Which severely limits the work-arounds one can try. 459658 11 techsoftadvanced 2011-08-31 09:14:05 -0700 (In reply to comment #10) > Just realized that having the "beforeload" event handler around isn't necessary to get broken behavior. It is enough that the event handler was registered at some point - removing it again won't help you any more, the "load" event for images will still fire too early. Which severely limits the work-arounds one can try. Wladimir Palant, I noticed that AdBlock Plus simply disables the majority of its checks for a hardcoded list of sites known to break because of this bug. Would it be instead possible to re-fire the load events for each element after a certain delay? I have yet to try this approach, but it seems plausible. I did some more research into how this breaks Google Mail and the AWS Management Console. It looks as though I was causing this bug to be more severe by triggering another bug. I can't give many more details than that, as I have yet to investigate this other bug, but the point is that the bug on which we are focusing can become even more severe when coupled with other bugs or even certain JavaScript techniques. For example, after adding a listener for beforeload, any modifications to an element's innerHTML property can have odd effects. 459660 12 trev 2011-08-31 09:17:25 -0700 (In reply to comment #11) > Would it be instead possible to re-fire the load events for each element after a certain delay? I tried that a while ago - it had no effect, the scripts in question only listen for the first "load" event (and it cannot be prevented). 459680 13 ap 2011-08-31 09:48:43 -0700 Also CC'ing Sam, who recently did some work on image load events. 464324 14 aestes 2011-09-08 13:50:41 -0700 I'm going to take a look at this today and see if I can make some progress. This bug has been assigned for almost a year but there doesn't seem to be any patch activity, so I assume I won't be stepping on anyone's feet. 464330 15 aestes 2011-09-08 13:53:00 -0700 (In reply to comment #12) > (In reply to comment #11) > > Would it be instead possible to re-fire the load events for each element after a certain delay? > > I tried that a while ago - it had no effect, the scripts in question only listen for the first "load" event (and it cannot be prevented). I believe this is due to <https://bugs.webkit.org/show_bug.cgi?id=33861>. We add events to the document's listener map on calls to addEventListener() but do not remove them on calls to removeEventListener(). 464334 16 aestes 2011-09-08 13:55:32 -0700 (In reply to comment #15) > I believe this is due to <https://bugs.webkit.org/show_bug.cgi?id=33861>. We add events to the document's listener map on calls to addEventListener() but do not remove them on calls to removeEventListener(). Sorry, this was meant to be in response to Comment #10. 464349 17 aestes 2011-09-08 14:12:07 -0700 I think this is due to this block of code starting at <http://trac.webkit.org/browser/trunk/Source/WebCore/loader/ImageLoader.cpp#L204>: if (newImage) { newImage->addClient(this); if (!m_element->document()->hasListenerType(Document::BEFORELOAD_LISTENER)) dispatchPendingBeforeLoadEvent(); else beforeLoadEventSender().dispatchEventSoon(this); } If newImage is cached, then the call to addClient() will cause the load event to fire *before* we dispatch the beforeload event. Not only is this the wrong order of events, but since dispatchPendingBeforeLoadEvent() is what ends up wiring up the cached image to the RenderImage, someone inspecting the image DOM object in an unload handler will see incorrect information. 464351 18 aestes 2011-09-08 14:13:10 -0700 s/unload/onload 464359 19 aestes 2011-09-08 14:17:14 -0700 In the case where Document::hasListenerType(Document::BEFORELOAD_LISTENER) returns false, things work correctly since the beforeload event is fired synchronously before the load event which is on a 0-delay timer. This bug is all about the case where hasListenerType() returns true and the beforeload event is also fired on a timer that happens to get queued behind the load event. 464410 20 aestes 2011-09-08 15:03:00 -0700 I think we're just calling setClient() too early. I'm testing a patch. 466187 21 aestes 2011-09-13 01:26:08 -0700 *** Bug 65388 has been marked as a duplicate of this bug. *** 466195 22 107152 aestes 2011-09-13 01:52:43 -0700 Created attachment 107152 Patch 466284 23 107152 mitz 2011-09-13 07:27:57 -0700 Comment on attachment 107152 Patch Sorry, I didn’t mean to r+ this. (Didn’t mean to r- either!). 466313 24 107152 darin 2011-09-13 08:35:35 -0700 Comment on attachment 107152 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=107152&action=review > Source/WebCore/loader/ImageLoader.cpp:209 > if (!m_element->document()->hasListenerType(Document::BEFORELOAD_LISTENER)) > dispatchPendingBeforeLoadEvent(); > else > beforeLoadEventSender().dispatchEventSoon(this); > + newImage->addClient(this); I think it’s worth adding a comment here. It’s probably not obvious to most readers that addClient often results in a load event dispatch and that the order here is important so that beforeload goes before load. A brief "why" comment could make that clear without making the code unpleasant. 468081 25 aestes 2011-09-15 14:20:10 -0700 Committed r95228: <http://trac.webkit.org/changeset/95228> 468184 26 webkit 2011-09-15 17:42:59 -0700 Superb! Thanks so much for your efforts, Andy. When can we expect this to be rolled out to the average web user? Thanks again, PaulH 468240 27 aestes 2011-09-15 19:14:09 -0700 Hi Paul, It's not really possible for me to answer your question. All I can say is that this fix is in WebKit trunk and can be tested in WebKit nightly builds. It's up to the downstream vendors of WebKit-based products, not the WebKit Open Source Project itself, to decide if and when a version of WebKit that includes this fix is shipped to customers. -Andy 496766 28 pkasting 2011-11-05 18:24:51 -0700 (In reply to comment #26) > When can we expect this to be rolled out to the average web user? I can't speak for other vendors, but for Chrome, this should be present in Chrome 16+, which currently corresponds to the Chrome beta, dev, and canary channels. If you ever want to check this sort of thing you can visit http://omahaproxy.appspot.com/viewer . Note the "base webkit revision" and compare to the revision number in comment 25. 520406 29 fam.lam 2011-12-13 15:13:23 -0800 Hi Andy, Thanks for fixing this! It unfortunately looks like a small part of this bug isn't fixed yet. I filed that in bug 74451. I'm not 100% sure about the relation to this issue, but it has a lot of similarities. Kind regards, Famlam 67292 2010-09-10 22:57:29 -0700 2010-09-10 22:57:29 -0700 A trivial Safari extension reproducing the bug test.safariextz application/x-safari-extension 5091 marc.hoyois eGFyIQAcAAEAAAAAAAAQEgAAAAAAAByqAAAAAXja1FbZsuI4En0ngn+4UfN4oxpvbB23qkPyjhfw jv3mDa/YGAPC/voRUHema7omOuZxHgikTCkzLemckx9/3I/12y0990XbfPtC/kZ8eUubuE2KJvv2 xbGFr6svf3yffNzD8/fJ28eljfHf20d8TsML3vH1UhzT7xRBEl+J9VeStIn57/Tid5r5mP285Lkp T+Oqvx7f+stQp9++9HlIfnl43j7aw6FPL9+Jj9mP0dPaF+Mj+MfsOXiEmH3GeM76ImvCy/Wcfv05 GU3MV8sFuWZ+Wz32/nrRTwE+SzIt8HNF1C9Lmi/+XdPbh5IOcnNo3/BJNv23L/nlcvp9NkMI/Ybo 39pzNqMIgpgR6xlekOCU/3ilePvYz4k1F17C1/Q1Z9PzpTgUcXhJv2uyLAQ2y0KeygCSIciwQV8c xTN5UsIYcYa/UdpAzm+xDgxegAZAWeDf2RFsYKa7EPg2qF17OtFMDfGGz7mGoXDgBGMJ9oEFN9FR u2oWQkr29Kkcq3/69hG96QMpOQWik5miSwXe/TadxKI7yMKmjkSBCL31NR4FXuNlEZAODxBaGnuI IrzDpdZDJJp06Jm1zJt1cnT7iIZ1XEALR8qnk0Ssb1GjZQblDnh8DD09f9oKCJO92eL8p0SqkJTH umYDpJc8pdnVoHP83Xva+KcNf92ntYStZrZIBL7CBvKxCHaDKhoQGEjxBFew643rjtXc4fi1xvWf NRMOJRz9/eY0nfxUadlmCp8EunOx9FEety5/kkUyj/GZheJ6HlH1aIjJxffqXj3qt8jmVQ1Uz6jT Ccw11nU1lGV8oQFCZK1OtOSI5gwe1+MAwMiQQ+DhV0CL79ZgiU13F0ubdLR2lGfardOuhjudSNaC VK4eV1mSnPTnwnS161m5dvPleXvnVoHXEd2ttzT2VIeX9igf44V5MWU+k4SiSQ5MtWNDQ47l6YTK h6hJKqkvFO6d4OiVZPi5dtu6CO1Ksdhcba4PKrvUQAMLDy23PKOTxMBQQuciUxOF0SfbIDnjV6Aw tczcuqW7MCR8daQwMoqdKEtJEteD7pQA8iYvM+TMJj3QWXO+lHrVzsvUbpha5Rg+6LWbr00na32g hLNjze711bItt/IM/UBaVdbfGPuaOvRxDiJer3VN4boLlViFndO6u+u3C72wzi59CnKBsfrp5MBc 6/diGwBDQ7WVexan5Mm7SaBF3FdmuQSZBgEQyyyTCowsSJdAe6BEMrH9sOJBCbAhFqcTfIPJBt/T DBm6Bvp/3Z1PGTzLgcNzl6XxIge8DFry5h5vusj3ffnWL5ZnyiQbwamtcjq5jGD7XG2snjkg0FjI ACQ9sGgSWwh9DFtqxm/4a7ehuIT2m8EILVUN3IV9g486OeRz0wlenUE8gyXLAkNG2WH157oEYLNw hX8I+bKCfAgNR8J8gbLI17JIf2Du5HtzXBPG9gNjRCjWoyzqddyYp+BYl/4eI1XczGVRuKYWRBgR BP7PY1onI8/Fa4XSp1yUYL9POdl0ElF+loh5LUtmG3hM5j+xjyN5mweDjI+VgSgMAQuJYL+5xAPM o6OR+dQa2+sXfwwQs0pQQDKmnP4v9Ui4VhzxUcdr51+4AsVHocTj8pHxwSoCEXhkHTXmqJaY6rLs DLMHN8YsMP0klEwi5tqbSid0MszzF9vNy4gibq/63dK3Vmg62Rl/OkmD5zEH3TQWI5nFzGE4AhwB zAuhTSQTbYvVLaL0MWZ/jviwYbSwJJ1Q5qBxMuKRr7w4OykD8Ij+fGfsg7t/wecc5gcIJJaplkia Tjq4akstp7uDK5KpbqeeHK6zUchOm/v7WaeXW+1mzvXOPVxpBjG0urucuISrd6eqCM1k1TB2demP M/wy78ghVwdknWzsvhg1EDPRlM+kB5FA3WumXDeZn4grTz7M+uR9lewaFAueMm+3emrsmj1bHFz2 Mp3wRHiYGxaXn/jbGmOSPKjDfC0ANoD1SpaqzX4Z55Rjl/J7yVDy9bQM6k0TKZWK1kE8nGv/MvTR GnNmiITuXFcXLysP7T3f7K2EcM/C4djlIpz1815Y5Kqqk2x7tQRnWFc9pHnaJ6pluAaCqaIGztOd 5E0nbMJKpR3Fe2sunvxu6OJVJzgcXRJZsDskAQlUR7Gp22jEWzcvZx+z/9Te/y7JvDxiCKLrpyRD 0QY6zKourwpxjYgHvQuAg4Vm9Ih9ia7Io43rjLytQf6HQLC5pnyKpYWBqZb8UWONp5e9a8mn72/k 8T6dCCVwXnKv2ZwjoFi811iibxFtZAbBvwSUy5BW8oRmZ6ReauRLQP2fbNOJV7KwLv8tYz9E7M6P wHxlaDW2+pGBr6/+wNxVDnQvX6/JxMuHhVRIbvHxXiVUXWFg8sE+wC3DGgV7OXOwwGJwnyJqPmqO gYxX88Fx3OUvzQemhj+1H39pPgbIYWLB4+DkUy+bzAskJiWcvSZS3DwBS34Cazr5E7TgD2jtAPeg 05Z9wkxrGVeJVLc7yxuulhf+eF0K78xhCM0bZ9ppP1/5DHy0LpSZnoy4vJTNu8MSrlvnSN0v+agX lNxmbit9wYudsB/C9VLcrbv33LEs05HFcOiooR2WQVCGgozFfecPylJdn8XNPgMLBA9BLAQrZpsH Dh1eic2lmzejK1TNKoiJ6L6PjjGZST51klMYlcWOYnvXbnp84likZ3a7g2Wi2KR5ul88Nm5sfese KssDg3iFvo5SlzZ90rDU9qymlpOnUnvf0GKoeKid3ZtDwy6w/NW1mhCKuY0ackwFbbkCS0sejnNL SGbeqbsuYk/3OGve0aNZNJuFtlUX+waER6ERIhXNUre9iZsDFoj+PaDTVeTPFlYQ9hg1PADhFp7x O+t+IY3A99FD7rDY2cCQZhA4uEXi4XQyG0HyWs3wQmY4BRvv2fBIiSKrLrfXxbpKgru7cdpljKRX hBLCDAktcJQ1PJlzk2IPSwKTDENcDSWiTYKcsQKD9Oy5+gDZFVJtcG7ZY7Zh/PrvhAJ/3VMqcJtL rQncAA4RBwyYtS8eoPQKYm6AQMZc8At+AFv87nhAlTKYTuredXCHs8zK5Hiw5i0Z78Sr1h9LvjDG u1aFVbj11zbeq4l0UvLIjt1Vqdlby7+Mxdw1tpoKFjN8d3yvNpwlMqTP7c5sdhsLarS7d1E0bHHh conN1hLP7laR2kdH8SIXqYBbwrlIjZ6gp6t3YrvxrbHE57Tcu6FE3pHk+nP+us85U6oKyxFviUcI /jwlhOO+2tZqlqoHUVnzCWSCljtJ44ZPtte+9G6LWg0O9HSSLm9eTgRsnl6snRUO/qIg+rjbrbWx 6Fe5xKjvoefvFtQpt5VWJQ+ic62Tatwf9pcgRnd95YZwm+O2k0lTGRAnsjn17UCcQCHiT0podVs4 5T2Y79/Z5Zbojta+0a7q4JKC8e3b/0Tt1we1L7b/onbcE/4fULuPdE4mn1TOadST2kcHUzuYf9oe 1P7Lqv8JAAD//7RZWW+jSBB+j5T/sJpXa4Y+OEfeSL5PMI5vXlbQ3dgYgx2DD/zrt7DjxJNJJitF KyGZLgoo99dVXd/HS9T/Neb7u8+i/izmKyOCefqUE1VDWPvtptL0miHvq4dpOo2jWlckZL3i5lgA vZvF4v6uMEoToy+2rWWpopbnhUK91EbUKD+1lwpimyiuPG5rvLLcFfpeJzH9laOo1c5jM8D7UX3J Tpnet4dMHkF54M12Y9rF07429nHB04NRdbea2MMxsvRJH7eO4259HPJEGVJjk4kTrqiH8aLvyHor rVRpph6gizWgPKRLHfr3jPZ13Y6H9NRB4bqRLAun7USppivRtLyu7Y3Urd+3q/PKoKLVFvVAQXh4 sAhpTZ5U6xiGHNj7mDX2XdQ4Fez9kpk0WqNj3V8Yposrk/4srvFGq1trWeXFqdeZe7XDpHd0y9WO 6XZ7dpMrKyYZ8TSa6Pd3A74gMWy6jlunKzNMXzlRXc25Bl++U5T7vxdl2CSgLL8tyreF9m2Z/ago wyp4xxtWwOOFZZVqcN7KYyuft+rS76xnmTcgcygPN32+OJfvpql2MwMKONtdevsRMBcj7ZLLqpyR 2t5sNau/cATYmgC7xVM5wXmjsQLeYQEvgsYCWI47Nd9wkNF8FodzYFYKcJUFcJecP50YHacO+N7f zSbW0pmWUc69cublnJnX6MybIHPO20r+pFkEWdK0cr8QmFcIb1uxCOesawf8aZmzLSV0n9smiCdq NccnZ3D4LR7WMPInKtc73zZI0Jw1yxnEBSPr8k5gWsDwdrxp7t5qFeXnvIStjVmbldmdBlSbZR61 LSjtkdQmKGaHoV7zF7NevaGExuPJz56ctLQMTqPEKQ1IV0NM2e9Q1LczurItKwjs/UqWyL5VLhyN 3qzXvb8bZdVhzxwc98rGrkR7qSMdnc1h1PbLvDRe1AS3ejTITO1RtcdeNgxUo0GZpduPohZn+3oq ejROH9kUOJJ1LLSW006bT52q4WxxpxW2jk26djObyfX5cZF6lUFhkOwXtYFdnvfa8tgYosjJKh2T bGl11rXpbgcz3pVWQ3eVhoeGeeISU8vZcaIaa7vV19xdbDpDbepYsVRe4KmzUSLFDLsascbHOFYX oy3z9o5VaZdbT/d3x4PO05CwaWS3BnJpOuo8dTDfhpJVapdP4sCno8UHG9XF/CITFqVnufGih74o mOehH6zEXwH/+9tVV80VILGt5MKnGD6rnmCO3XidCLaOefIgE0M2VI0YalG6tV88z1IpNjT0HWE4 hgj9PB9F6SqiQhDvvqXIbrRh/TuWh1j/SbSfMnWKEnu9O/qDW/Tq5v6iNCN5iOSflIInuLmvbvPt erd5SFLX94vSZXCxB/ys5ea/Z8MuEduHyN2yxTpbB0lROhsul8AHyjCYrs7RmosHpCkKhJSfno1c 7AMm4vUDobKuY0RgAl9sZ48gzp2pjAjOFePL8HwlzTbigQdbwdL1NoPJzMd3F2jgr6QiSX8kru9u A3FMRZyL8zk613/5gjO5Ssn8RkZeiXieLh6wBu98Pn++chW18ys3qvZV11axrN8I22C+fgq46uPu ZrPK1ybEIx2/z0/B5pt09c3nMtgL/v2tzj9olmA9coIZ4lzIqqp5RKUq4Z5MKcOq4bmyUHxFdxHG AuB8+6CXaI7p1mXpx6/wie8JD8sIuYy7yECGSwnDzFc9XzaIjzHBAmlF6fcnXZPtdSIvIOVTfYPP M0A/qoN/BgCduEVFyn3fAET/ABBk3gcAyQp5FyCMdOV/AwhR1ZOJkIUna6qBhCoMzGTZpQKpridj olAMc8q+AJAhqC+wp3guxy4WqiYI4Yj6vk+wTjWfGZBHDOtfByivkD9gKpL0M4TkPyCk6h8ABPn+ fgaR/w0eX6cEkoQbjGpcQ4QpmGNN+IwjOU8ezVdcnbGvwMOByPqKhwxuuLIifM/3uaoLjQoiwKxq KkCk5sX/q/CcC9wyeReblzO4Lf/cWJTOHx//BQAA//8DAL03BRxCy6qnvuGJZ7HNJ/8kysprcNRm Azkk4s5gcZyHH+gjzIaneuh8retZCUQCTh222/rtvlznPFcUB0QGlWCVgjCZ0rW4C6RTu0eJd+26 1uw14vpCISoaJztZepM7vN/3iIaixq3cFzs9/4lW8iVev+5h3rtQeUDFx7PABTCpwkQ/4KzW8M97 KcbT6bsq5cx/vY84GwzkLksP7hcvf321Ia7rPGIjvRX/dBVP93D76sNn2BinOywv4bhcufpjUNAW 2joCyXGOlZYuShhwOvBWqPDTmYb4FaowfpP33U7oXSLhmrApkkCMVAFmPHM9dbmHF6jgWq1O5UoT hYXAuwoMs11ajZKEXAeTW3Pk6Eqbaj9xtjbra0J42u2YMQrDMBAEb2UVgjQqXarJA/IDYewX5AMp 8oX0enowtxCRKKlN2AEzAls2SKC7tZlhedwvZtnMkrntbEMSrw9CNwbfIYQQQohjA9b3k5ZCCDE4 Hwpd6eYG7wc6vucA+NTdlW5u8LlARzrRmS50pZubhxYYPsAvgwkFTCEodNU+CjFicuW9/m/f878Q 4p/rfFyv6/IKBMNG4NaNfzUBwX8Wzt1cNQJCHIwnSKsKaXjajVPBbptAED3XX7Hlbja+VFVFiBzj qJZIioqdqsftMo63Xe+i3SEOf9/Bxg7GpuoJNO+9mXlvILp722r2Cs4ra26DSXgTMDDSFsq83Aar 5cP4c3AXj6KPybfZ8mc2Z6VWHlm2uk8XMxaMOZ+WpQbOk2XCsnSRLxn14Hz+FLBgg1h+4Xy324Wi YYXSbhui55mzJTisU2o2JkFYYBHQmEP3s3WoWiiJ8ehD9AfqeFrhxrqIN+9U8uho0/hROMm+2toq H/G21gpmD/eVKTQkypda1E9iC301gsch2aIAg2qt4GImmQk3+5HhPxuYtU3IABkSrn4+WOv3+hTe DOlzsoutLN8z+uLJkHRg2CV/42wnlSZufsKsQUrgDKTnAcylUyX6I/iOtjAKd1JSTTgnDq9n0Ye/ u0cjjJ+IET/ennc/ggT8fvIVc9dOMX8jC00S/3WLyeUtMnBb5Ruyvx7ED/jlFQKbSgl+MI+Fkboq gOUgKwcsEy/gO/Ggq4C/s1N4Bd2B25WmWnfTGkhoVRaC9nkURq0pEbb6nvZ99k22HoZpxwkR3/+m 8egvfideIHja40rJTy7NTc0r0UtMSXEtAzJ8MotLUvNSizSUklLT8otSc/ITU5R0FNJK85JLMvPz NDSra3UUSopKUzWtuQDOYRX9 76914 2010-12-17 15:06:55 -0800 2010-12-17 15:06:55 -0800 Minimal testcase beforeload_bug.html text/html 3630 trev PCFET0NUWVBFIGh0bWw+Cgo8aHRtbD4KPGhlYWQ+CiAgPHRpdGxlPmJlZm9yZWxvYWQgZXZlbnQg aGFuZGxlciBpbmZsdWVuY2luZyBpbWFnZSBsb2FkPC90aXRsZT4KCiAgPHNjcmlwdCB0eXBlPSJ0 ZXh0L2phdmFzY3JpcHQiPgogICAgZG9jdW1lbnQuYWRkRXZlbnRMaXN0ZW5lcigiYmVmb3JlbG9h ZCIsIGZ1bmN0aW9uKGUpCiAgICB7CiAgICAgIGNvbnNvbGUubG9nKCJkb2N1bWVudCdzIGJlZm9y ZWxvYWQgZXZlbnQsIGltYWdlIHdpZHRoID0gIiArIGUudGFyZ2V0LndpZHRoKTsKICAgIH0sIHRy dWUpOwoKICAJZnVuY3Rpb24gcnVuVGVzdCgpCiAgICB7CiAgCQl2YXIgaW1nID0gZG9jdW1lbnQu Y3JlYXRlRWxlbWVudCgiaW1nIik7CiAgCQlpbWcub25sb2FkID0gZnVuY3Rpb24oKQogICAgICB7 CiAgCQkJY29uc29sZS5sb2coImltYWdlJ3MgbG9hZCBldmVudCwgaW1hZ2Ugd2lkdGggPSAiKyB0 aGlzLndpZHRoKTsKICAJCX0KICAJCWRvY3VtZW50LmJvZHkuYXBwZW5kQ2hpbGQoaW1nKTsKICAJ CWltZy5zcmMgPSAiZGF0YTppbWFnZS9wbmc7YmFzZTY0LGlWQk9SdzBLR2dvQUFBQU5TVWhFVWdB QUFHUUFBQUJrQ0FNQUFBQkhQR1ZtQUFBQ3RWQk1WRVglMkYlMkZ3QUFBQUQlMkIlMkZnQUJBUUQz OXdEYjJ3QU5EUUQwOUFEOSUyRlFEdDdRQXNMQUFLQ2dEUzBnRDglMkZBRE96Z0NscFFEMjlnQ1Fr QUFtSmdENiUyQmdENyUyQndDb3FBQVNFZ0I2ZWdBQ0FnQ09qZ0FIQndEZDNRQmhZUUFMQ3dBSkNR QlBUd0RtNWdEYzNBQURBd0NGaFFBdkx3RHI2d0FPRGdDUGp3RFQwd0IzZHdDVWxBRGg0UUIlMkJm Z0RsNVFEajR3QVVGQUJ0YlFCTFN3QyUyQnZnRHk4Z0QxOVFBRkJRQmZYd0FkSFFETXpBQVJFUUMz dHdCOWZRQUVCQUIlMkZmd0JRVUFEdzhBQ3NyQUJWVlFEZTNnQVRFd0RIeHdBJTJGUHdDNnVnQ3pz d0IyZGdBZUhnREZ4UUExTlFETHl3QU1EQUFHQmdBcEtRQ25wd0ElMkJQZ0N3c0FCaVlnQkNRZ0JZ V0FCMGRBQ3BxUUFYRndEbzZBQnNiQUQ1JTJCUUFXRmdDTmpRQmxaUUNnb0FEejh3QnZid0N4c1FE cDZRQnViZ0RpNGdDV2xnQ2twQUM0dUFEUjBRQWZId0RKeVFCblp3QkRRd0JkWFFBSUNBRFUxQUJH UmdBUUVBQVlHQUI3ZXdER3hnQkhSd0JiV3dCS1NnRFcxZ0J4Y1FEazVBRFkyQUJNVEFDaG9RQmdZ QUJxYWdBYUdnRHE2Z0NMaXdDRGd3Qk5UUUN5c2dCSlNRQ1NrZ0RhMmdDOXZRQndjQUQ0JTJCQUE3 T3dBWkdRQnJhd0F1TGdDbXBnQkJRUURFeEFDMnRnQjRlQUNabVFEeDhRQnBhUUE0T0FBMk5nQjhm QUFWRlFBckt3QXRMUUJhV2dDVmxRRFoyUURYMXdCWFZ3QWtKQURnNEFDdnJ3QTNOd0FxS2dEQndR QmpZd0RBd0FDQ2dnRGYzd0RWMVFBb0tBQ2RuUUMxdFFDZm53QWpJd0NNakFBME5BQThQQUMwdEFC VVZBQWNIQUJrWkFEbjV3QWhJUUNSa1FBNk9nQ2FtZ0NLaWdCMWRRQzh2QUJUVXdDZW5nQmNYQUR2 N3dDSWlBQTlQUUJlWGdDRWhBQk9UZ0FQRHdESXlBQXdNQUFuSndDJTJGdndDdXJnQXlNZ0RDd2dC RVJBQWJHd0FsSlFCWldRQnljZ0JJU0FCRlJRQ1ltQUJXVmdCemN3Q3FxZ0REd3dEUHp3QjVlUUJv YUFES3lnQWlJZ0N0clFETnpRQiUyQlc0bTVBQUFFNWtsRVFWUjRYdTNhVTdQc1dCakg0ZmRObTl1 MmJkdTJlV3pidG0zYkdOdTJiYyUyRm5tSzNPU1hheWtuUjNjamUlMkYyJTJGJTJGRlU2bFVyVlNx RnFDMFVtNnRXajdwMkEwMWpPU3hydEZyJTJCNmFxTFNxVWxCVEV1JTJGT05uZE04Z1JQMWJuclJo b055SUV2VDRtZFFRRzZyUzljeUo1SElvakFRTFRUbVdTZVFUaGMxU0VwZlhlenZHTkxXVG9IMDZr Tjg3VWVNVDFyQXZwNDRZQ2NTRkZjQ2RqY3dZZEFlcE9FT09OVEQxWklSMVZvZE9KaCUyQlc1TTBa SEFXT05HOXg2UWd4bHh3cW9RRjRraFZFamlaN3BJWUVwSUJUbWVaTG95ODVRcHkxQzJFaEtoQm52 YVJrYW9Na0NuOUJ5U2tKUlJreTNVSlA1SmlCUmtMdmMySGhCOEJXWnVuNVVHbWdNeWQ1Q0pmWnNx TlVNJTJCUFIwejFJSHNKVzhZaGNhQkFHOW1JZTRZU2lPZEtGcklSRkdsZUlRTjVWUSUyRksxUEVJ OGEwaEEyRmViakd0VjRIYmpOSlglMkZHYVpSWkNId1RUeVBwRklQdlF4RHRYVUVRYnNvbXBUaG9m Z3JzWEN5bndhOFNJWlp2cGIlMkJuSWxheWh2c0EzTHFnVVJhJTJCRVlZdlFnR0VudVNKZkRmQmJE TTBoWFlSVlVRc2FRSWlDMEdSbVZNWVpFNXJDQUVrTGFSNUhvY3NJZUVNUkVUQnA2bURiSUhBcFhD Q0VsZFNQSUx0TGVpcXpjNkNHZlBSU0FVSEVqeUV6UzdNZEdVdW5oY2Zhd1NSQzVvMExBdkNTSlNE OEp5UlpFZEU4aDRGSGlmSVNOdkU0UHg5bkRDUkJzQ2dKMkU5Y3JpNWlHdjVrZVFyMlpneXBkR0ds RlVEMUhuZ09aeUE3R1VNc2NEbGhFanBabXFJZ2l6dyUyRnFrTzQ4OCUyRnpvdTRCMDJub1F6dFVk YmxtQVhPVmMlMkJ2U1l6Um9XdDlpR3ZGOUJySVZ3U1hEWFhJc2VKa3pGWm1DWEhEanlYdnhYVm9K b2ZyQWNoTnV6ZTNycTAyYmc5bURtbFA3M0drRkNVMkVTS0o0VjZwVkhYb0lFNVJGWFVDdVA2QlVF JTJGdSUyRiUyRlBKUW5LRWhTSGxGRG8lMkZKSUZIZ3BqOFRDZmVXUmZNaFdIb21ITmtweFpENGNM RkhhOERnTXZ1bEtJNXBnd0VTbGtac0l1RnBweEFjQmV3M0tHZ09IRVJCZGxFVnl3NGVSTW1XUkNC eEd6azlUMG5DOU1JSmdETzlLUmR5ekIxaFRjSXAzMkl1anlCbUtiNTJBMnBsNnlVYnBDJTJGZ2k3 eEE0aG9TJTJGQ2R6eXN4RHhvem5TaUlUTmpEOVhWaHJUR0lJN3VPT2NwVGpjMGdnSkx5enp0VG9j em5jcWQxdU9Oc1IlMkZCV2UwUnVKbzF5JTJCTE1Ca3VEVGhhRGhlWjRVMGptQWFjREhkeHJMcjda aklSc051SVkyMEk0TTRuOEJFU1BodTQ5YmpibU9DMHQ5ZFlnTnVObm5meWJFUnZETVUxMWdjeEVE enJBZHo2VXJWb2E5R1pjejI1ZmZ0dGs2ZGh6czBQbDBUVGMzalhldUJwSVRJUjlBTyUyQnJOZVlm OWxaM3NhUWlkMCUyQlByV255ODcyTmlPakJiT0JyJTJCUElSaXF1QUclMkJmck1wRGtZTFRkZ0p2 aHBaeENDN1JBMzhhbjA5VlNPNnpRN2xBYUMyT1IzQWJrQnI0UExXaGlRJTJGSXVoMzR4V1FnVmFQ aUlrRmZBVGxQemNuVFgxJTJGWCUyQnRLQWFXNWJvSjkxUDVBciUyRndhNUNINmJETUs1SmglMkZ6 dWh4UlVMRGRiZGJlZ01sNkVFelhpWHdJSHRXQmJPa25JaiUyQkMzM25LaGlRaUNjRiUyQnVRdzNK Q1A0dlR4R3ZFb0l3WDBXMlo2RGpHQzIwMiUyQmZTa1F4Qkg5WTU1d3hlU0tLSTJoYzRZeXg1MGVV Z21EMFQ0NGJVM01rWDElMkY0T2N3eFl1c3ZLcFNNWU00a2l3Tkd6VVU3cjVUcyUyQnMxZUlqYmIl MkZzc3hXWGQlMkZ0NGNJJTJCRU9MWklSYzBKJTJGenBCSlh6NlU0ZkdHcHVialVJQzdvcWp1MHps MjltaHVYZmtwSVVQOFZjVkdHUzJTcXY2ZTd4R2J5QVFPYWYzd2lDeEdkUSUyQmo4M1JkT2FQODNO Rk0lMkZDbERxcU5qOCUyQlBtUkpwVFVmNEZBMXJEQ2lQbTdBQUFBQUVsRlRrU3VRbUNDIjsKICAJ fQogIDwvc2NyaXB0PgoKPC9oZWFkPgo8Ym9keT4KCjxwPkNsaWNrIHRoZSBidXR0b24gdG8gbG9h ZCB0aGUgaW1hZ2UuIFJlc3VsdHMgZnJvbSBsb2FkIGFuZCBiZWZvcmVsb2FkIGV2ZW50cyB3aWxs IGJlIHByaW50ZWQgdG8gY29uc29sZS4gTm90ZSB0aGF0IHRoZSBidWcgb25seSBhcHBlYXJzIG9u IHRoZSBzZWNvbmQgcnVuLjwvcD4KCjxidXR0b24gb25jbGljaz0icnVuVGVzdCgpIj5DbGljayB0 byBsb2FkIGltYWdlPC9idXR0b24+Cgo8L2JvZHk+CjwvaHRtbD4K 107152 2011-09-13 01:52:43 -0700 2011-09-13 08:35:35 -0700 Patch bug-45586-20110913015247.patch text/plain 5912 aestes U3VidmVyc2lvbiBSZXZpc2lvbjogOTQ5ODAKZGlmZiAtLWdpdCBhL0xheW91dFRlc3RzL0NoYW5n ZUxvZyBiL0xheW91dFRlc3RzL0NoYW5nZUxvZwppbmRleCAwYTQ0OTRhODNhYzc1YzE4OWYzNzQ5 MDRhYmJkMTk2ZjQ3MTFjYjY2Li4xMDJkMDg5YTA0YWRkNmUzNjIyYjFjOWI2MGM3OTUzMWRhM2Fk MjA1IDEwMDY0NAotLS0gYS9MYXlvdXRUZXN0cy9DaGFuZ2VMb2cKKysrIGIvTGF5b3V0VGVzdHMv Q2hhbmdlTG9nCkBAIC0xLDMgKzEsMTMgQEAKKzIwMTEtMDktMTMgIEFuZHkgRXN0ZXMgIDxhZXN0 ZXNAYXBwbGUuY29tPgorCisgICAgICAgIEhhdmluZyBhbiBlbXB0eSBsaXN0ZW5lciB0byBiZWZv cmVsb2FkIGV2ZW50cyBjaGFuZ2VzIHRoZSBiZWhhdmlvciBvZiBvdGhlciBzY3JpcHRzCisgICAg ICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD00NTU4NgorCisgICAg ICAgIFJldmlld2VkIGJ5IE5PQk9EWSAoT09QUyEpLgorCisgICAgICAgICogZmFzdC9kb20vYmVm b3JlbG9hZC9jYWNoZWQtaW1hZ2UtYmVmb3JlLWxvYWQtZXhwZWN0ZWQudHh0OiBBZGRlZC4KKyAg ICAgICAgKiBmYXN0L2RvbS9iZWZvcmVsb2FkL2NhY2hlZC1pbWFnZS1iZWZvcmUtbG9hZC5odG1s OiBBZGRlZC4KKwogMjAxMS0wOS0xMiAgQmV0aCBEYWtpbiAgPGJkYWtpbkBhcHBsZS5jb20+CiAK ICAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTY3ODk4CmRp ZmYgLS1naXQgYS9MYXlvdXRUZXN0cy9mYXN0L2RvbS9iZWZvcmVsb2FkL2NhY2hlZC1pbWFnZS1i ZWZvcmUtbG9hZC1leHBlY3RlZC50eHQgYi9MYXlvdXRUZXN0cy9mYXN0L2RvbS9iZWZvcmVsb2Fk L2NhY2hlZC1pbWFnZS1iZWZvcmUtbG9hZC1leHBlY3RlZC50eHQKbmV3IGZpbGUgbW9kZSAxMDA2 NDQKaW5kZXggMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMC4uOWEwNzQ5 Y2Q3MGM3YTRiNjU5ZmFkMzhkMGVhNzZiZTFjZjgwMTg3ZAotLS0gL2Rldi9udWxsCisrKyBiL0xh eW91dFRlc3RzL2Zhc3QvZG9tL2JlZm9yZWxvYWQvY2FjaGVkLWltYWdlLWJlZm9yZS1sb2FkLWV4 cGVjdGVkLnR4dApAQCAtMCwwICsxLDUgQEAKK1RoaXMgaXMgYSByZWdyZXNzaW9uIHRlc3QgZm9y IGh0dHA6Ly93ZWJraXQub3JnL2IvNDU1ODYuIEl0IHZlcmlmaWVzIHRoYXQgdGhlIGJlZm9yZWxv YWQgZXZlbnQgZmlyZXMgYmVmb3JlIHRoZSBsb2FkIGV2ZW50IHdoZW4gbG9hZGluZyBhIGNhY2hl ZCBpbWFnZSwgYW5kIHRoYXQgdGhlIGltYWdlJ3Mgc3R5bGUgYXR0cmlidXRlcyBjYW4gYmUgcXVl cmllZCBpbiB0aGUgbG9hZCBldmVudCBoYW5kbGVyLiBPbiBzdWNjZXNzLCB5b3Ugc2hvdWxkIHNl ZSB0d28gbGluZXMgb2YgZ3JlZW4gdGV4dCBzdGFydGluZyB3aXRoIHRoZSB3b3JkICdQQVNTJy4K KworUEFTUzogbG9hZCBmaXJlZCBhZnRlciBiZWZvcmVsb2FkIGZvciBhIGNhY2hlZCBpbWFnZS4K K1BBU1M6IGltYWdlIGlzIDE2cHggd2lkZS4KKyAKZGlmZiAtLWdpdCBhL0xheW91dFRlc3RzL2Zh c3QvZG9tL2JlZm9yZWxvYWQvY2FjaGVkLWltYWdlLWJlZm9yZS1sb2FkLmh0bWwgYi9MYXlvdXRU ZXN0cy9mYXN0L2RvbS9iZWZvcmVsb2FkL2NhY2hlZC1pbWFnZS1iZWZvcmUtbG9hZC5odG1sCm5l dyBmaWxlIG1vZGUgMTAwNjQ0CmluZGV4IDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAw MDAwMDAwMDAuLjMzNWVlOGZlYjQzMzg0ZGFiODAwZDdlOTI4NTlhMTNmNDUwNjljNDcKLS0tIC9k ZXYvbnVsbAorKysgYi9MYXlvdXRUZXN0cy9mYXN0L2RvbS9iZWZvcmVsb2FkL2NhY2hlZC1pbWFn ZS1iZWZvcmUtbG9hZC5odG1sCkBAIC0wLDAgKzEsMzggQEAKKzxzY3JpcHQgc3JjPSJyZXNvdXJj ZXMvcHJpbnQuanMiPjwvc2NyaXB0PgorPHNjcmlwdD4KKyAgICBpZiAod2luZG93LmxheW91dFRl c3RDb250cm9sbGVyKSB7CisgICAgICAgIGxheW91dFRlc3RDb250cm9sbGVyLmR1bXBBc1RleHQo KTsKKyAgICAgICAgbGF5b3V0VGVzdENvbnRyb2xsZXIud2FpdFVudGlsRG9uZSgpOworICAgIH0K KworICAgIHdpbmRvdy5fYmVmb3JlbG9hZGZpcmVkID0gZmFsc2U7CisKKyAgICBmdW5jdGlvbiBy dW5UZXN0KCkKKyAgICB7CisgICAgICAgIGRvY3VtZW50LmFkZEV2ZW50TGlzdGVuZXIoImJlZm9y ZWxvYWQiLCBmdW5jdGlvbihldmVudCkgeworICAgICAgICAgICAgd2luZG93Ll9iZWZvcmVsb2Fk ZmlyZWQgPSB0cnVlOworICAgICAgICB9LCB0cnVlKTsKKworICAgICAgICBjYWNoZWRJbWFnZSA9 IG5ldyBJbWFnZSgpOworICAgICAgICBkb2N1bWVudC5ib2R5LmFwcGVuZENoaWxkKGNhY2hlZElt YWdlKTsKKyAgICAgICAgY2FjaGVkSW1hZ2Uub25sb2FkID0gZnVuY3Rpb24oZXZlbnQpIHsKKyAg ICAgICAgICAgIGlmICh3aW5kb3cuX2JlZm9yZWxvYWRmaXJlZCkKKyAgICAgICAgICAgICAgICBw cmludCgiUEFTUzogbG9hZCBmaXJlZCBhZnRlciBiZWZvcmVsb2FkIGZvciBhIGNhY2hlZCBpbWFn ZS4iLCAiZ3JlZW4iKTsKKyAgICAgICAgICAgIGVsc2UKKyAgICAgICAgICAgICAgICBwcmludCgi RkFJTDogbG9hZCBmaXJlZCBiZWZvcmUgYmVmb3JlbG9hZCBmb3IgYSBjYWNoZWQgaW1hZ2UuIiwg InJlZCIpOworICAgICAgICAgICAgICAgIAorICAgICAgICAgICAgaWYgKHRoaXMud2lkdGggPT0g MTYpCisgICAgICAgICAgICAgICAgcHJpbnQoIlBBU1M6IGltYWdlIGlzIDE2cHggd2lkZS4iLCAi Z3JlZW4iKTsKKyAgICAgICAgICAgIGVsc2UKKyAgICAgICAgICAgICAgICBwcmludCgiRkFJTDog aW1hZ2UgY2xhaW1zIG5vdCB0byBiZSAxNnB4IHdpZGUuIiwgInJlZCIpOworCisgICAgICAgICAg ICBpZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKQorICAgICAgICAgICAgICAgIGxheW91 dFRlc3RDb250cm9sbGVyLm5vdGlmeURvbmUoKTsKKyAgICAgICAgfQorICAgICAgICBjYWNoZWRJ bWFnZS5zcmMgPSBkb2N1bWVudC5nZXRFbGVtZW50QnlJZCgib3JpZ2luYWxJbWFnZSIpLnNyYzsK KyAgICAgICAgY2FjaGVkSW1hZ2Uuc3R5bGUudmlzaWJpbGl0eSA9ICJoaWRkZW4iOworICAgIH0K Kzwvc2NyaXB0PgorPHA+VGhpcyBpcyBhIHJlZ3Jlc3Npb24gdGVzdCBmb3IgPGEgaHJlZj0iaHR0 cDovL3dlYmtpdC5vcmcvYi80NTU4NiI+aHR0cDovL3dlYmtpdC5vcmcvYi80NTU4NjwvYT4uIEl0 IHZlcmlmaWVzIHRoYXQgdGhlIGJlZm9yZWxvYWQgZXZlbnQgZmlyZXMgYmVmb3JlIHRoZSBsb2Fk IGV2ZW50IHdoZW4gbG9hZGluZyBhIGNhY2hlZCBpbWFnZSwgYW5kIHRoYXQgdGhlIGltYWdlJ3Mg c3R5bGUgYXR0cmlidXRlcyBjYW4gYmUgcXVlcmllZCBpbiB0aGUgbG9hZCBldmVudCBoYW5kbGVy LiBPbiBzdWNjZXNzLCB5b3Ugc2hvdWxkIHNlZSB0d28gbGluZXMgb2YgZ3JlZW4gdGV4dCBzdGFy dGluZyB3aXRoIHRoZSB3b3JkICdQQVNTJy48L3A+Cis8ZGl2IGlkPSJjb25zb2xlIj48L2Rpdj4K KzxpbWcgaWQ9Im9yaWdpbmFsSW1hZ2UiIHN0eWxlPSJ2aXNpYmlsaXR5OmhpZGRlbiIgb25sb2Fk PSJydW5UZXN0KCkiIHNyYz0iLi4vLi4vaW1hZ2VzL3Jlc291cmNlcy90ZXN0LWxvYWQuanBnIj4K ZGlmZiAtLWdpdCBhL1NvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZyBiL1NvdXJjZS9XZWJDb3JlL0No YW5nZUxvZwppbmRleCBkNGQ3MjI4Y2I0YzMyODFiNzE2ZTMzNDM3NjJjMzBjNTZlNGMzMGZmLi4x OTQxYjIzMmYzNTcyMTRlMGRhZjk1YTBlYjBkYTRhNzI1MTFjYWQ2IDEwMDY0NAotLS0gYS9Tb3Vy Y2UvV2ViQ29yZS9DaGFuZ2VMb2cKKysrIGIvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCkBAIC0x LDMgKzEsMjggQEAKKzIwMTEtMDktMTMgIEFuZHkgRXN0ZXMgIDxhZXN0ZXNAYXBwbGUuY29tPgor CisgICAgICAgIEhhdmluZyBhbiBlbXB0eSBsaXN0ZW5lciB0byBiZWZvcmVsb2FkIGV2ZW50cyBj aGFuZ2VzIHRoZSBiZWhhdmlvciBvZiBvdGhlciBzY3JpcHRzCisgICAgICAgIGh0dHBzOi8vYnVn cy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD00NTU4NgorCisgICAgICAgIFJldmlld2VkIGJ5 IE5PQk9EWSAoT09QUyEpLgorCisgICAgICAgIFRlc3Q6IGZhc3QvZG9tL2JlZm9yZWxvYWQvY2Fj aGVkLWltYWdlLWJlZm9yZS1sb2FkLmh0bWwKKyAgICAgICAgCisgICAgICAgIFdoZW4gbG9hZGlu ZyBhIGNhY2hlZCBpbWFnZSBhZnRlciBhIGJlZm9yZWxvYWQgaGFuZGxlciBoYXMgYmVlbgorICAg ICAgICBpbnN0YWxsZWQgb24gdGhlIGRvY3VtZW50LCBJbWFnZUxvYWRlciB3b3VsZCBkaXNwYXRj aCBib3RoIHRoZQorICAgICAgICBiZWZvcmVsb2FkIGFuZCBsb2FkIGV2ZW50cyBhc3luY2hyb25v dXNseSBpbiBzdWNoIGEgd2F5IHRoYXQgY2F1c2VkCisgICAgICAgIGxvYWQgdG8gZmlyZSBmaXJz dC4gU2luY2UgYSBzaWRlIGVmZmVjdCBvZiBmaXJpbmcgdGhlIGJlZm9yZWxvYWQgZXZlbnQKKyAg ICAgICAgaXMgdG8gd2lyZSB1cCB0aGUgQ2FjaGVkSW1hZ2UgdG8gaXRzIGFzc29jaWF0ZWQgUmVu ZGVySW1hZ2Ugb2JqZWN0LAorICAgICAgICB0aGlzIHdvcmsgd2FzIG5vdCBkb25lIGJ5IHRoZSB0 aW1lIGxvYWQgZmlyZWQsIGFuZCBzY3JpcHRzIHRoYXQgcXVlcmllZAorICAgICAgICByZW5kZXJl ci1kZXBlbmRlbnQgYXR0cmlidXRlcyBvZiB0aGUgaW1hZ2UgaW4gYW4gb25sb2FkIGhhbmRsZXIg d291bGQKKyAgICAgICAgZ2V0IGJvZ3VzIHZhbHVlcyBpbiByZXR1cm4uCisKKyAgICAgICAgRml4 IHRoaXMgYnkgZW5zdXJpbmcgbG9hZCBmaXJlcyBhZnRlciBiZWZvcmVsb2FkIGluIHRoZSBjYWNo ZWQgaW1hZ2UgY2FzZS4KKworICAgICAgICAqIGxvYWRlci9JbWFnZUxvYWRlci5jcHA6CisgICAg ICAgIChXZWJDb3JlOjpJbWFnZUxvYWRlcjo6dXBkYXRlRnJvbUVsZW1lbnQpOiBDYWxsIHNldENs aWVudCgpIGFmdGVyCisgICAgICAgIGRpc3BhdGNoaW5nIGJlZm9yZWxvYWQsIHNpbmNlIHNldENs aWVudCgpIHdpbGwgZGlzcGF0Y2ggdGhlIGxvYWQgZXZlbnQKKyAgICAgICAgaWYgdGhlIGltYWdl IGlzIGNhY2hlZC4KKwogMjAxMS0wOS0xMiAgQmV0aCBEYWtpbiAgPGJkYWtpbkBhcHBsZS5jb20+ CiAKICAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTY3ODk4 CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9sb2FkZXIvSW1hZ2VMb2FkZXIuY3BwIGIvU291 cmNlL1dlYkNvcmUvbG9hZGVyL0ltYWdlTG9hZGVyLmNwcAppbmRleCBjYzc0NzQ5ZDA2NGMwNGM0 ODc3ZTU4M2RjNWYwNTQ1YzJlMmRmZDUxLi4yYjE1OGNiNmY3ZWYwMzc5NDMzMTczOWEzYmMyNjdl OTgxYWM3ZWQ5IDEwMDY0NAotLS0gYS9Tb3VyY2UvV2ViQ29yZS9sb2FkZXIvSW1hZ2VMb2FkZXIu Y3BwCisrKyBiL1NvdXJjZS9XZWJDb3JlL2xvYWRlci9JbWFnZUxvYWRlci5jcHAKQEAgLTIwMiwx MSArMjAyLDExIEBAIHZvaWQgSW1hZ2VMb2FkZXI6OnVwZGF0ZUZyb21FbGVtZW50KCkKICAgICAg ICAgbV9pbWFnZUNvbXBsZXRlID0gIW5ld0ltYWdlOwogCiAgICAgICAgIGlmIChuZXdJbWFnZSkg ewotICAgICAgICAgICAgbmV3SW1hZ2UtPmFkZENsaWVudCh0aGlzKTsKICAgICAgICAgICAgIGlm ICghbV9lbGVtZW50LT5kb2N1bWVudCgpLT5oYXNMaXN0ZW5lclR5cGUoRG9jdW1lbnQ6OkJFRk9S RUxPQURfTElTVEVORVIpKQogICAgICAgICAgICAgICAgIGRpc3BhdGNoUGVuZGluZ0JlZm9yZUxv YWRFdmVudCgpOwogICAgICAgICAgICAgZWxzZQogICAgICAgICAgICAgICAgIGJlZm9yZUxvYWRF dmVudFNlbmRlcigpLmRpc3BhdGNoRXZlbnRTb29uKHRoaXMpOworICAgICAgICAgICAgbmV3SW1h Z2UtPmFkZENsaWVudCh0aGlzKTsKICAgICAgICAgfQogICAgICAgICBpZiAob2xkSW1hZ2UpCiAg ICAgICAgICAgICBvbGRJbWFnZS0+cmVtb3ZlQ2xpZW50KHRoaXMpOwo=