111005 2013-02-27 13:04:52 -0800 Underline should round to match other content. 2013-02-28 19:41:51 -0800 1 1 1 Unclassified WebKit Platform 528+ (Nightly build) All All RESOLVED FIXED P2 Normal --- 1 bungeman bungeman eric junov leviw senorblanco webkit.review.bot oldest_to_newest 843409 0 bungeman 2013-02-27 13:04:52 -0800 Underline should round to match other content. 843414 1 190582 bungeman 2013-02-27 13:07:33 -0800 Created attachment 190582 Patch 843565 2 190618 bungeman 2013-02-27 15:30:48 -0800 Created attachment 190618 Patch 843572 3 bungeman 2013-02-27 15:44:22 -0800 With http://trac.webkit.org/changeset/117528 the old subpixel placement of underlines was replaced with a snapping version. However, that implementation takes the floor of the y value, while all other content uses the rounded y value. This means that the underlines do not move in step with other content, including text. See crbug.com/173065 for more explanation. This change simply simply adds 0.5 to the y value in order to round the same as other content. It appears that one image test is affected. In the future this should use a valid implementation of roundToDevicePixels as the current implementation (and this newer one) cause the underline to snap to css pixels instead of physical pixels. As a result, on HiDPI devices the underline is snapped to even pixels and only moves once while the text (drawn to device pixels) moves twice as it is slowly moved vertically. 844060 4 bungeman 2013-02-28 07:19:50 -0800 Adding those involved in the 117528 change, as they'll probably be more familiar with this. 844076 5 190618 senorblanco 2013-02-28 07:52:38 -0800 Comment on attachment 190618 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190618&action=review > Source/WebCore/ChangeLog:8 > + No new tests (OOPS!). You should mention the affected tests here (e.g., decorations-with-text-combine). 844155 6 190743 bungeman 2013-02-28 09:45:34 -0800 Created attachment 190743 Patch 844176 7 190743 leviw 2013-02-28 10:08:26 -0800 Comment on attachment 190743 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > Source/WebCore/platform/graphics/skia/GraphicsContextSkia.cpp:754 > + // Round to nearest pixel to match text and other content. > + r.fTop = WebCoreFloatToSkScalar(floorf(pt.y() + 0.5f)); Do we ever have negative values here? LayoutUnit rounds differently around zero: http://trac.webkit.org/browser/trunk/Source/WebCore/platform/LayoutUnit.h#L225 844189 8 bungeman 2013-02-28 10:21:56 -0800 (In reply to comment #7) > (From update of attachment 190743 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > > > Source/WebCore/platform/graphics/skia/GraphicsContextSkia.cpp:754 > > + // Round to nearest pixel to match text and other content. > > + r.fTop = WebCoreFloatToSkScalar(floorf(pt.y() + 0.5f)); > > Do we ever have negative values here? LayoutUnit rounds differently around zero: http://trac.webkit.org/browser/trunk/Source/WebCore/platform/LayoutUnit.h#L225 Skia currently always rounds X.5 up (towards +inf) when rounding text positions. As a result, this change should always keep the line in-step with the text. If the content rounds differently, it already rounds differently from the text. This may require some follow-up investigation, as noted in Comment #3 about roundToDevicePixels being implemented. 844190 9 190743 senorblanco 2013-02-28 10:23:25 -0800 Comment on attachment 190743 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review >>> Source/WebCore/platform/graphics/skia/GraphicsContextSkia.cpp:754 >>> + r.fTop = WebCoreFloatToSkScalar(floorf(pt.y() + 0.5f)); >> >> Do we ever have negative values here? LayoutUnit rounds differently around zero: http://trac.webkit.org/browser/trunk/Source/WebCore/platform/LayoutUnit.h#L225 > > Skia currently always rounds X.5 up (towards +inf) when rounding text positions. As a result, this change should always keep the line in-step with the text. If the content rounds differently, it already rounds differently from the text. This may require some follow-up investigation, as noted in Comment #3 about roundToDevicePixels being implemented. Nit: floating point constants don't get a trailing 'f' in WebKit code. 844220 10 190743 leviw 2013-02-28 10:47:33 -0800 Comment on attachment 190743 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > Source/WebCore/ChangeLog:8 > + Test: fast/text/decorations-with-text-combine.html So this test already covers this? Is that why you're not adding one? I'd feel more comfortable if you had test expectations for at least one platform with this patch so the difference could be seen. 844253 11 bungeman 2013-02-28 11:20:30 -0800 (In reply to comment #9) > (From update of attachment 190743 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > > >>> Source/WebCore/platform/graphics/skia/GraphicsContextSkia.cpp:754 > >>> + r.fTop = WebCoreFloatToSkScalar(floorf(pt.y() + 0.5f)); > >> > >> Do we ever have negative values here? LayoutUnit rounds differently around zero: http://trac.webkit.org/browser/trunk/Source/WebCore/platform/LayoutUnit.h#L225 > > > > Skia currently always rounds X.5 up (towards +inf) when rounding text positions. As a result, this change should always keep the line in-step with the text. If the content rounds differently, it already rounds differently from the text. This may require some follow-up investigation, as noted in Comment #3 about roundToDevicePixels being implemented. > > Nit: floating point constants don't get a trailing 'f' in WebKit code. Indeed, this silly thing is in the style guide. Why make calculations be done with doubles only to then convert them back to float? There is no equivalent to CGFloat in WebKit, and everything is float, so why not force 'f' instead? I have already gone through all stages of grief except acceptance. Do I not get a pass since GraphicsContext::rotate(float) in this file also uses 'f'? 844256 12 bungeman 2013-02-28 11:23:46 -0800 (In reply to comment #10) > (From update of attachment 190743 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > > > Source/WebCore/ChangeLog:8 > > + Test: fast/text/decorations-with-text-combine.html > > So this test already covers this? Is that why you're not adding one? I'd feel more comfortable if you had test expectations for at least one platform with this patch so the difference could be seen. This test is the only one which changes. The effect of this change is to move one of the vertical lines in this test one pixel to the right. Every other test which uses an underline is also covering this, but I would prefer not to list all of those. None of them seem to have changed. 844263 13 190763 bungeman 2013-02-28 11:28:05 -0800 Created attachment 190763 Patch 844268 14 leviw 2013-02-28 11:29:56 -0800 (In reply to comment #12) > (In reply to comment #10) > > (From update of attachment 190743 [details] [details]) > > View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > > > > > Source/WebCore/ChangeLog:8 > > > + Test: fast/text/decorations-with-text-combine.html > > > > So this test already covers this? Is that why you're not adding one? I'd feel more comfortable if you had test expectations for at least one platform with this patch so the difference could be seen. > > This test is the only one which changes. The effect of this change is to move one of the vertical lines in this test one pixel to the right. Every other test which uses an underline is also covering this, but I would prefer not to list all of those. None of them seem to have changed. If the rest don't change, they're not covering this :-/ Ideally, we'd have a simple test that would make clear the regression if this change were ever broken. Why aren't you bundling the updated test expectation for the platform you're using with this patch? 844271 15 190763 senorblanco 2013-02-28 11:32:02 -0800 Comment on attachment 190763 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190763&action=review > Source/WebCore/platform/graphics/skia/GraphicsContextSkia.cpp:754 > + r.fTop = WebCoreFloatToSkScalar(floorf(pt.y() + static_cast<float>(0.5))); Heh, cute... Actually, reading this rule again, it says "*Unless required in order to force floating point math*, do not append .0, .f and .0f to floating point literals." Since you do actually want to force floating point math here, I think f is fine. Sorry for the runaround.. 844328 16 bungeman 2013-02-28 12:01:59 -0800 (In reply to comment #14) > (In reply to comment #12) > > (In reply to comment #10) > > > (From update of attachment 190743 [details] [details] [details]) > > > View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > > > > > > > Source/WebCore/ChangeLog:8 > > > > + Test: fast/text/decorations-with-text-combine.html > > > > > > So this test already covers this? Is that why you're not adding one? I'd feel more comfortable if you had test expectations for at least one platform with this patch so the difference could be seen. > > > > This test is the only one which changes. The effect of this change is to move one of the vertical lines in this test one pixel to the right. Every other test which uses an underline is also covering this, but I would prefer not to list all of those. None of them seem to have changed. > > If the rest don't change, they're not covering this :-/ They are, there shouldn't be any change, but because of the silly way in which vertical text is implemented in WebKit (horizontal layout but rotated) this one changed. It wouldn't be changing if vertical text layout were sane. > > Ideally, we'd have a simple test that would make clear the regression if this change were ever broken. Why aren't you bundling the updated test expectation for the platform you're using with this patch? Because the change is very, very minor and I'm on precise, so it's very difficult to get a new baseline locally. Beside that, I would never trust any locally produced baseline, there are too many things which can go wrong with that. If you have a usable means of generating baselines ahead of time, I'm all ears. A while ago I was fighting to make the results of layout tests retrievable from try bots, but I'm not sure where that ended up. Also, this is going to change again anyway (the text part at least), since there is a related Skia change which will go in with the next mass Skia re-baseline. I'm also against adding new pixel tests, if it can be at all avoided. Also, as mentioned in Comment #3, this is still known to be wrong, just less wrong. roundToDevicePixels needs to be implemented and then used here. 844333 17 190743 senorblanco 2013-02-28 12:08:37 -0800 Comment on attachment 190743 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review OK. Let's go with this version. r=me >>>>> Source/WebCore/platform/graphics/skia/GraphicsContextSkia.cpp:754 >>>>> + r.fTop = WebCoreFloatToSkScalar(floorf(pt.y() + 0.5f)); >>>> >>>> Do we ever have negative values here? LayoutUnit rounds differently around zero: http://trac.webkit.org/browser/trunk/Source/WebCore/platform/LayoutUnit.h#L225 >>> >>> Skia currently always rounds X.5 up (towards +inf) when rounding text positions. As a result, this change should always keep the line in-step with the text. If the content rounds differently, it already rounds differently from the text. This may require some follow-up investigation, as noted in Comment #3 about roundToDevicePixels being implemented. >> >> Nit: floating point constants don't get a trailing 'f' in WebKit code. > > Indeed, this silly thing is in the style guide. Why make calculations be done with doubles only to then convert them back to float? There is no equivalent to CGFloat in WebKit, and everything is float, so why not force 'f' instead? I have already gone through all stages of grief except acceptance. Do I not get a pass since GraphicsContext::rotate(float) in this file also uses 'f'? Sure. 844409 18 leviw 2013-02-28 13:19:18 -0800 (In reply to comment #16) > (In reply to comment #14) > > (In reply to comment #12) > > > (In reply to comment #10) > > > > (From update of attachment 190743 [details] [details] [details] [details]) > > > > View in context: https://bugs.webkit.org/attachment.cgi?id=190743&action=review > > > > > > > > > Source/WebCore/ChangeLog:8 > > > > > + Test: fast/text/decorations-with-text-combine.html > > > > > > > > So this test already covers this? Is that why you're not adding one? I'd feel more comfortable if you had test expectations for at least one platform with this patch so the difference could be seen. > > > > > > This test is the only one which changes. The effect of this change is to move one of the vertical lines in this test one pixel to the right. Every other test which uses an underline is also covering this, but I would prefer not to list all of those. None of them seem to have changed. > > > > If the rest don't change, they're not covering this :-/ > They are, there shouldn't be any change, but because of the silly way in which vertical text is implemented in WebKit (horizontal layout but rotated) this one changed. It wouldn't be changing if vertical text layout were sane. I really don't understand this argument. I do understand that this doesn't affect the way the majority of tests with underlines render, but it's still testable. Absent the updated expectations, it's impossible for me to verify this is correct, and ignoring the one test that covers it on all platforms means it lands without coverage. Also for the record, I'm running precise and pass the pixel tests for that tests locally. > > > > Ideally, we'd have a simple test that would make clear the regression if this change were ever broken. Why aren't you bundling the updated test expectation for the platform you're using with this patch? > Because the change is very, very minor and I'm on precise, so it's very difficult to get a new baseline locally. Beside that, I would never trust any locally produced baseline, there are too many things which can go wrong with that. If you have a usable means of generating baselines ahead of time, I'm all ears. A while ago I was fighting to make the results of layout tests retrievable from try bots, but I'm not sure where that ended up. > > Also, this is going to change again anyway (the text part at least), since there is a related Skia change which will go in with the next mass Skia re-baseline. I'm also against adding new pixel tests, if it can be at all avoided. Also, as mentioned in Comment #3, this is still known to be wrong, just less wrong. roundToDevicePixels needs to be implemented and then used here. 844461 19 leviw 2013-02-28 14:06:02 -0800 To be clear, I defer to Stephen on this. You'll need to uncheck your previous patch as obsolete to land it with his R+ CQ+. 844505 20 190743 webkit.review.bot 2013-02-28 14:56:09 -0800 Comment on attachment 190743 Patch Clearing flags on attachment: 190743 Committed r144375: <http://trac.webkit.org/changeset/144375> 844506 21 webkit.review.bot 2013-02-28 14:56:14 -0800 All reviewed patches have been landed. Closing bug. 844735 22 bungeman 2013-02-28 19:41:51 -0800 (In reply to comment #21) > All reviewed patches have been landed. Closing bug. Re-baseline and removal of line from TestExpectations with 144380 and 144396. 190582 2013-02-27 13:07:33 -0800 2013-02-27 15:30:45 -0800 Patch bug-111005-20130227160346.patch text/plain 1462 bungeman SW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDE0NDIzMCkKKysrIFNvdXJjZS9XZWJDb3JlL0NoYW5n ZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDE1IEBACisyMDEzLTAyLTI3ICBCZW4gV2Fn bmVyICA8YnVuZ2VtYW5AY2hyb21pdW0ub3JnPgorCisgICAgICAgIFVuZGVybGluZSBzaG91bGQg cm91bmQgdG8gbWF0Y2ggb3RoZXIgY29udGVudC4KKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtp dC5vcmcvc2hvd19idWcuY2dpP2lkPTExMTAwNQorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9E WSAoT09QUyEpLgorCisgICAgICAgIE5vIG5ldyB0ZXN0cyAoT09QUyEpLgorCisgICAgICAgICog cGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9HcmFwaGljc0NvbnRleHRTa2lhLmNwcDoKKyAgICAgICAg KFdlYkNvcmU6OkdyYXBoaWNzQ29udGV4dDo6ZHJhd0xpbmVGb3JUZXh0KToKKwogMjAxMy0wMi0y NyAgTGlhbmdodWkgQ2hlbiAgPGxpYWNoZW5AcmltLmNvbT4KIAogICAgICAgICBbQmxhY2tCZXJy eV0gVXNlciBjcmVkZW50aWFscyBpcyBub3QgY29ycmVjdGx5IGhhbmRsZWQKSW5kZXg6IFNvdXJj ZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL3NraWEvR3JhcGhpY3NDb250ZXh0U2tpYS5jcHAK PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PQotLS0gU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9HcmFw aGljc0NvbnRleHRTa2lhLmNwcAkocmV2aXNpb24gMTQ0MDc3KQorKysgU291cmNlL1dlYkNvcmUv cGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9HcmFwaGljc0NvbnRleHRTa2lhLmNwcAkod29ya2luZyBj b3B5KQpAQCAtNzUwLDcgKzc1MCw4IEBAIHZvaWQgR3JhcGhpY3NDb250ZXh0OjpkcmF3TGluZUZv clRleHQoY28KICAgICBTa1JlY3QgcjsKICAgICByLmZMZWZ0ID0gV2ViQ29yZUZsb2F0VG9Ta1Nj YWxhcihwdC54KCkpOwogICAgIC8vIEF2b2lkIGFudGktYWxpYXNpbmcgbGluZXMuIEN1cnJlbnRs eSwgdGhlc2UgYXJlIGFsd2F5cyBob3Jpem9udGFsLgotICAgIHIuZlRvcCA9IFdlYkNvcmVGbG9h dFRvU2tTY2FsYXIoZmxvb3JmKHB0LnkoKSkpOworICAgIC8vIFJvdW5kIHRvIG5lYXJlc3QgcGl4 ZWwgdG8gbWF0Y2ggdGV4dCBhbmQgb3RoZXIgY29udGVudC4KKyAgICByLmZUb3AgPSBXZWJDb3Jl RmxvYXRUb1NrU2NhbGFyKGZsb29yZihwdC55KCkgKyAwLjVmKSk7CiAgICAgci5mUmlnaHQgPSBy LmZMZWZ0ICsgV2ViQ29yZUZsb2F0VG9Ta1NjYWxhcih3aWR0aCk7CiAgICAgci5mQm90dG9tID0g ci5mVG9wICsgU2tJbnRUb1NjYWxhcih0aGlja25lc3MpOwogCg== 190618 2013-02-27 15:30:48 -0800 2013-02-28 09:45:30 -0800 Patch bug-111005-20130227182701.patch text/plain 2189 bungeman SW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDE0NDIzMCkKKysrIFNvdXJjZS9XZWJDb3JlL0NoYW5n ZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDE1IEBACisyMDEzLTAyLTI3ICBCZW4gV2Fn bmVyICA8YnVuZ2VtYW5AY2hyb21pdW0ub3JnPgorCisgICAgICAgIFVuZGVybGluZSBzaG91bGQg cm91bmQgdG8gbWF0Y2ggb3RoZXIgY29udGVudC4KKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtp dC5vcmcvc2hvd19idWcuY2dpP2lkPTExMTAwNQorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9E WSAoT09QUyEpLgorCisgICAgICAgIE5vIG5ldyB0ZXN0cyAoT09QUyEpLgorCisgICAgICAgICog cGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9HcmFwaGljc0NvbnRleHRTa2lhLmNwcDoKKyAgICAgICAg KFdlYkNvcmU6OkdyYXBoaWNzQ29udGV4dDo6ZHJhd0xpbmVGb3JUZXh0KToKKwogMjAxMy0wMi0y NyAgTGlhbmdodWkgQ2hlbiAgPGxpYWNoZW5AcmltLmNvbT4KIAogICAgICAgICBbQmxhY2tCZXJy eV0gVXNlciBjcmVkZW50aWFscyBpcyBub3QgY29ycmVjdGx5IGhhbmRsZWQKSW5kZXg6IFNvdXJj ZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL3NraWEvR3JhcGhpY3NDb250ZXh0U2tpYS5jcHAK PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PQotLS0gU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9HcmFw aGljc0NvbnRleHRTa2lhLmNwcAkocmV2aXNpb24gMTQ0MDc3KQorKysgU291cmNlL1dlYkNvcmUv cGxhdGZvcm0vZ3JhcGhpY3Mvc2tpYS9HcmFwaGljc0NvbnRleHRTa2lhLmNwcAkod29ya2luZyBj b3B5KQpAQCAtNzUwLDcgKzc1MCw4IEBAIHZvaWQgR3JhcGhpY3NDb250ZXh0OjpkcmF3TGluZUZv clRleHQoY28KICAgICBTa1JlY3QgcjsKICAgICByLmZMZWZ0ID0gV2ViQ29yZUZsb2F0VG9Ta1Nj YWxhcihwdC54KCkpOwogICAgIC8vIEF2b2lkIGFudGktYWxpYXNpbmcgbGluZXMuIEN1cnJlbnRs eSwgdGhlc2UgYXJlIGFsd2F5cyBob3Jpem9udGFsLgotICAgIHIuZlRvcCA9IFdlYkNvcmVGbG9h dFRvU2tTY2FsYXIoZmxvb3JmKHB0LnkoKSkpOworICAgIC8vIFJvdW5kIHRvIG5lYXJlc3QgcGl4 ZWwgdG8gbWF0Y2ggdGV4dCBhbmQgb3RoZXIgY29udGVudC4KKyAgICByLmZUb3AgPSBXZWJDb3Jl RmxvYXRUb1NrU2NhbGFyKGZsb29yZihwdC55KCkgKyAwLjVmKSk7CiAgICAgci5mUmlnaHQgPSBy LmZMZWZ0ICsgV2ViQ29yZUZsb2F0VG9Ta1NjYWxhcih3aWR0aCk7CiAgICAgci5mQm90dG9tID0g ci5mVG9wICsgU2tJbnRUb1NjYWxhcih0aGlja25lc3MpOwogCkluZGV4OiBMYXlvdXRUZXN0cy9w bGF0Zm9ybS9jaHJvbWl1bS9UZXN0RXhwZWN0YXRpb25zCj09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIExheW91dFRl c3RzL3BsYXRmb3JtL2Nocm9taXVtL1Rlc3RFeHBlY3RhdGlvbnMJKHJldmlzaW9uIDE0NDA3NykK KysrIExheW91dFRlc3RzL3BsYXRmb3JtL2Nocm9taXVtL1Rlc3RFeHBlY3RhdGlvbnMJKHdvcmtp bmcgY29weSkKQEAgLTEyOSw2ICsxMjksOSBAQCB3ZWJraXQub3JnL2IvMTEwMzI0IFsgRGVidWcg XSBmYXN0L2pzL3JlCiAjIFRFTVBPUkFSSUxZIFNLSVBQRUQgVEVTVFMKICMgLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIAor IyBVbmRlcmxpbmUgcGxhY2VtZW50IG5lZWRzIHRvIGJlIHJvdW5kZWQgdG8gbWF0Y2ggb3RoZXIg Y29udGVudC4KK3dlYmtpdC5vcmcvYi8xMTEwMDUgZmFzdC90ZXh0L2RlY29yYXRpb25zLXdpdGgt dGV4dC1jb21iaW5lLmh0bWwgWyBJbWFnZU9ubHlGYWlsdXJlIF0KKwogIyBDb250ZXh0IG1lbnUg YmVoYXZpb3IgZGlmZmVycyBvbiBNYWMuIFRoaXMgdGVzdCBuZWVkcyB0byBiZSBhZGp1c3RlZCBm b3IgTWFjLgogd2Via2l0Lm9yZy9iLzEwMjk4NyBbIE1hYyBdIGZhc3QvZXZlbnRzL2NvbnRleHQt bm9kcmFnLmh0bWwgWyBTa2lwIF0KIAo= 190743 2013-02-28 09:45:34 -0800 2013-02-28 14:56:09 -0800 Patch bug-111005-20130228124146.patch text/plain 2287 bungeman SW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDE0NDIzMCkKKysrIFNvdXJjZS9XZWJDb3JlL0NoYW5n ZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDE2IEBACisyMDEzLTAyLTI3ICBCZW4gV2Fn bmVyICA8YnVuZ2VtYW5AY2hyb21pdW0ub3JnPgorCisgICAgICAgIFVuZGVybGluZSBzaG91bGQg cm91bmQgdG8gbWF0Y2ggb3RoZXIgY29udGVudC4KKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtp dC5vcmcvc2hvd19idWcuY2dpP2lkPTExMTAwNQorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9E WSAoT09QUyEpLgorCisgICAgICAgIFRlc3Q6IGZhc3QvdGV4dC9kZWNvcmF0aW9ucy13aXRoLXRl eHQtY29tYmluZS5odG1sCisKKyAgICAgICAgKiBwbGF0Zm9ybS9ncmFwaGljcy9za2lhL0dyYXBo aWNzQ29udGV4dFNraWEuY3BwOgorICAgICAgICAoV2ViQ29yZTo6R3JhcGhpY3NDb250ZXh0Ojpk cmF3TGluZUZvclRleHQpOgorICAgICAgICBSb3VuZCAoaW5zdGVhZCBvZiBmbG9vcikgdW5kZXJs aW5lcyB0byBtYXRjaCBvdGhlciBjb250ZW50LgorCiAyMDEzLTAyLTI3ICBMaWFuZ2h1aSBDaGVu ICA8bGlhY2hlbkByaW0uY29tPgogCiAgICAgICAgIFtCbGFja0JlcnJ5XSBVc2VyIGNyZWRlbnRp YWxzIGlzIG5vdCBjb3JyZWN0bHkgaGFuZGxlZApJbmRleDogU291cmNlL1dlYkNvcmUvcGxhdGZv cm0vZ3JhcGhpY3Mvc2tpYS9HcmFwaGljc0NvbnRleHRTa2lhLmNwcAo9PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBT b3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9za2lhL0dyYXBoaWNzQ29udGV4dFNraWEu Y3BwCShyZXZpc2lvbiAxNDQwNzcpCisrKyBTb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGlj cy9za2lhL0dyYXBoaWNzQ29udGV4dFNraWEuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC03NTAsNyAr NzUwLDggQEAgdm9pZCBHcmFwaGljc0NvbnRleHQ6OmRyYXdMaW5lRm9yVGV4dChjbwogICAgIFNr UmVjdCByOwogICAgIHIuZkxlZnQgPSBXZWJDb3JlRmxvYXRUb1NrU2NhbGFyKHB0LngoKSk7CiAg ICAgLy8gQXZvaWQgYW50aS1hbGlhc2luZyBsaW5lcy4gQ3VycmVudGx5LCB0aGVzZSBhcmUgYWx3 YXlzIGhvcml6b250YWwuCi0gICAgci5mVG9wID0gV2ViQ29yZUZsb2F0VG9Ta1NjYWxhcihmbG9v cmYocHQueSgpKSk7CisgICAgLy8gUm91bmQgdG8gbmVhcmVzdCBwaXhlbCB0byBtYXRjaCB0ZXh0 IGFuZCBvdGhlciBjb250ZW50LgorICAgIHIuZlRvcCA9IFdlYkNvcmVGbG9hdFRvU2tTY2FsYXIo Zmxvb3JmKHB0LnkoKSArIDAuNWYpKTsKICAgICByLmZSaWdodCA9IHIuZkxlZnQgKyBXZWJDb3Jl RmxvYXRUb1NrU2NhbGFyKHdpZHRoKTsKICAgICByLmZCb3R0b20gPSByLmZUb3AgKyBTa0ludFRv U2NhbGFyKHRoaWNrbmVzcyk7CiAKSW5kZXg6IExheW91dFRlc3RzL3BsYXRmb3JtL2Nocm9taXVt L1Rlc3RFeHBlY3RhdGlvbnMKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVzdHMvcGxhdGZvcm0vY2hy b21pdW0vVGVzdEV4cGVjdGF0aW9ucwkocmV2aXNpb24gMTQ0MDc3KQorKysgTGF5b3V0VGVzdHMv cGxhdGZvcm0vY2hyb21pdW0vVGVzdEV4cGVjdGF0aW9ucwkod29ya2luZyBjb3B5KQpAQCAtMTI5 LDYgKzEyOSw5IEBAIHdlYmtpdC5vcmcvYi8xMTAzMjQgWyBEZWJ1ZyBdIGZhc3QvanMvcmUKICMg VEVNUE9SQVJJTFkgU0tJUFBFRCBURVNUUwogIyAtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQogCisjIFVuZGVybGluZSBwbGFj ZW1lbnQgbmVlZHMgdG8gYmUgcm91bmRlZCB0byBtYXRjaCBvdGhlciBjb250ZW50Lgord2Via2l0 Lm9yZy9iLzExMTAwNSBmYXN0L3RleHQvZGVjb3JhdGlvbnMtd2l0aC10ZXh0LWNvbWJpbmUuaHRt bCBbIEltYWdlT25seUZhaWx1cmUgXQorCiAjIENvbnRleHQgbWVudSBiZWhhdmlvciBkaWZmZXJz IG9uIE1hYy4gVGhpcyB0ZXN0IG5lZWRzIHRvIGJlIGFkanVzdGVkIGZvciBNYWMuCiB3ZWJraXQu b3JnL2IvMTAyOTg3IFsgTWFjIF0gZmFzdC9ldmVudHMvY29udGV4dC1ub2RyYWcuaHRtbCBbIFNr aXAgXQogCg== 190763 2013-02-28 11:28:05 -0800 2013-02-28 14:30:19 -0800 Patch bug-111005-20130228142417.patch text/plain 2306 bungeman SW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDE0NDIzMCkKKysrIFNvdXJjZS9XZWJDb3JlL0NoYW5n ZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDE2IEBACisyMDEzLTAyLTI3ICBCZW4gV2Fn bmVyICA8YnVuZ2VtYW5AY2hyb21pdW0ub3JnPgorCisgICAgICAgIFVuZGVybGluZSBzaG91bGQg cm91bmQgdG8gbWF0Y2ggb3RoZXIgY29udGVudC4KKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtp dC5vcmcvc2hvd19idWcuY2dpP2lkPTExMTAwNQorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9E WSAoT09QUyEpLgorCisgICAgICAgIFRlc3Q6IGZhc3QvdGV4dC9kZWNvcmF0aW9ucy13aXRoLXRl eHQtY29tYmluZS5odG1sCisKKyAgICAgICAgKiBwbGF0Zm9ybS9ncmFwaGljcy9za2lhL0dyYXBo aWNzQ29udGV4dFNraWEuY3BwOgorICAgICAgICAoV2ViQ29yZTo6R3JhcGhpY3NDb250ZXh0Ojpk cmF3TGluZUZvclRleHQpOgorICAgICAgICBSb3VuZCAoaW5zdGVhZCBvZiBmbG9vcikgdW5kZXJs aW5lcyB0byBtYXRjaCBvdGhlciBjb250ZW50LgorCiAyMDEzLTAyLTI3ICBMaWFuZ2h1aSBDaGVu ICA8bGlhY2hlbkByaW0uY29tPgogCiAgICAgICAgIFtCbGFja0JlcnJ5XSBVc2VyIGNyZWRlbnRp YWxzIGlzIG5vdCBjb3JyZWN0bHkgaGFuZGxlZApJbmRleDogU291cmNlL1dlYkNvcmUvcGxhdGZv cm0vZ3JhcGhpY3Mvc2tpYS9HcmFwaGljc0NvbnRleHRTa2lhLmNwcAo9PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBT b3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9za2lhL0dyYXBoaWNzQ29udGV4dFNraWEu Y3BwCShyZXZpc2lvbiAxNDQwNzcpCisrKyBTb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGlj cy9za2lhL0dyYXBoaWNzQ29udGV4dFNraWEuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC03NTAsNyAr NzUwLDggQEAgdm9pZCBHcmFwaGljc0NvbnRleHQ6OmRyYXdMaW5lRm9yVGV4dChjbwogICAgIFNr UmVjdCByOwogICAgIHIuZkxlZnQgPSBXZWJDb3JlRmxvYXRUb1NrU2NhbGFyKHB0LngoKSk7CiAg ICAgLy8gQXZvaWQgYW50aS1hbGlhc2luZyBsaW5lcy4gQ3VycmVudGx5LCB0aGVzZSBhcmUgYWx3 YXlzIGhvcml6b250YWwuCi0gICAgci5mVG9wID0gV2ViQ29yZUZsb2F0VG9Ta1NjYWxhcihmbG9v cmYocHQueSgpKSk7CisgICAgLy8gUm91bmQgdG8gbmVhcmVzdCBwaXhlbCB0byBtYXRjaCB0ZXh0 IGFuZCBvdGhlciBjb250ZW50LgorICAgIHIuZlRvcCA9IFdlYkNvcmVGbG9hdFRvU2tTY2FsYXIo Zmxvb3JmKHB0LnkoKSArIHN0YXRpY19jYXN0PGZsb2F0PigwLjUpKSk7CiAgICAgci5mUmlnaHQg PSByLmZMZWZ0ICsgV2ViQ29yZUZsb2F0VG9Ta1NjYWxhcih3aWR0aCk7CiAgICAgci5mQm90dG9t ID0gci5mVG9wICsgU2tJbnRUb1NjYWxhcih0aGlja25lc3MpOwogCkluZGV4OiBMYXlvdXRUZXN0 cy9wbGF0Zm9ybS9jaHJvbWl1bS9UZXN0RXhwZWN0YXRpb25zCj09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIExheW91 dFRlc3RzL3BsYXRmb3JtL2Nocm9taXVtL1Rlc3RFeHBlY3RhdGlvbnMJKHJldmlzaW9uIDE0NDA3 NykKKysrIExheW91dFRlc3RzL3BsYXRmb3JtL2Nocm9taXVtL1Rlc3RFeHBlY3RhdGlvbnMJKHdv cmtpbmcgY29weSkKQEAgLTEyOSw2ICsxMjksOSBAQCB3ZWJraXQub3JnL2IvMTEwMzI0IFsgRGVi dWcgXSBmYXN0L2pzL3JlCiAjIFRFTVBPUkFSSUxZIFNLSVBQRUQgVEVTVFMKICMgLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0K IAorIyBVbmRlcmxpbmUgcGxhY2VtZW50IG5lZWRzIHRvIGJlIHJvdW5kZWQgdG8gbWF0Y2ggb3Ro ZXIgY29udGVudC4KK3dlYmtpdC5vcmcvYi8xMTEwMDUgZmFzdC90ZXh0L2RlY29yYXRpb25zLXdp dGgtdGV4dC1jb21iaW5lLmh0bWwgWyBJbWFnZU9ubHlGYWlsdXJlIF0KKwogIyBDb250ZXh0IG1l bnUgYmVoYXZpb3IgZGlmZmVycyBvbiBNYWMuIFRoaXMgdGVzdCBuZWVkcyB0byBiZSBhZGp1c3Rl ZCBmb3IgTWFjLgogd2Via2l0Lm9yZy9iLzEwMjk4NyBbIE1hYyBdIGZhc3QvZXZlbnRzL2NvbnRl eHQtbm9kcmFnLmh0bWwgWyBTa2lwIF0KIAo=