76932 2012-01-24 12:45:27 -0800 [Qt] soft hyphen does not break the line, instead causes content to overlap 2012-10-22 05:10:01 -0700 1 1 1 Unclassified WebKit Layout and Rendering 528+ (Nightly build) Unspecified Unspecified CLOSED WONTFIX P2 Normal --- 88168 0 andreas_nordal_4 dtharp allan.jensen hausmann menard pierre.rossi tomz oldest_to_newest 541571 0 123786 andreas_nordal_4 2012-01-24 12:45:27 -0800 Created attachment 123786 testcase demonstrating overlapping text caused by soft hyphen Using a browser with QtWebKit (tested: Konqueror, Rekonq, Arora, QtTestBrowser), this html is not displayed correctly when the browser window is narrow: <table border="1"> <tr> <td>aaaaaaaaaaaaaa&shy;bbbbbbbbbbbbbb</td> <td>something</td> </tr> </table> Actual Results: 1) WebKit does not break the line. 2) Worse, it apparently thinks it has broken the line, allowing that "something" to the right to overlap with the "bbbbbbbbbbbbbb". 3) The overlapping text is unreadable. Expected Results: The soft hyphen should appear as a hyphen at the end of the "aaaaaaaaaaaaaa", and "bbbbbbbbbbbbbb" should move to the next line (out of the way for that "something" to the right). Non-Webkit browsers work fine. I have not tested non-QtWebKit WebKit browsers. The bug was first submitted here: https://bugs.kde.org/show_bug.cgi?id=289182 Version numbers (How to infer the version of WebKit?): libQtWebKit4 4.7.4+2.2.0-2.1.2 kwebkitpart 1.2.0git20111019-1.2 542499 1 ap 2012-01-25 11:47:30 -0800 This works for me in Safari on Mac. 576032 2 allan.jensen 2012-03-12 02:50:27 -0700 Works with QtWebKit from trunks (using MiniBrowser --desktop). Confirmed broken with QtWebKit from Qt 4.7.4. 577804 3 dtharp 2012-03-13 14:14:46 -0700 This bug is related to bug 27593. The root of the issue is the same: QTextBoundaryFinder::QTextBoundaryFinder() does not respect soft hyphen as a line break. The manifestation in this case is that the line inside the table cell never gets broken at the soft hyphen, causing a collision. This differs from the problem cited in bug 27593, namely that the hyphen is never drawn but lines break properly. Having just run that test, I'd disagree that the lines break properly at all: the lines simply continue to break on word boundaries, not respecting soft hyphen at all. 578076 4 allan.jensen 2012-03-13 18:27:25 -0700 The bug also appears fixed in Qt 4.8.0 To summarize my tests (all on Linux): QtWebkit from Qt 4.7.4 : Bug confirmed QtWebkit from 4.8.0: Bug fixed (all soft hyphens works as expected) QtWebkit from trunk: Bug fixed (all soft hyphens works as expected) Should this bug be closed as fixed, or does it still occur in recent Qt versions on some platforms? Btw. The same test results counts for bug 27593 579908 5 132152 dtharp 2012-03-15 16:57:40 -0700 Created attachment 132152 Patch 580584 6 dtharp 2012-03-16 09:36:44 -0700 Note that this patch appears to fix bug 27593 as well. I hesitate to mark as duplicate because the symptoms differ, but the root cause is the same. 580610 7 dtharp 2012-03-16 09:58:57 -0700 Also note that I am using the latest available 4.8 SDK from Nokia (http://qt.nokia.com/downloads ). Bug 27593 (recently closed) mentions that nokia has fixed the soft hyphen problem in their implementation of QTextBoundaryFinder, but this fix does not appear to be widely released at this point. The patch provided here is therefore an interim solution until Nokia releases and updated SDK. 581846 8 132152 menard 2012-03-19 09:11:50 -0700 Comment on attachment 132152 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=132152&action=review Is the issue reported in Qt? > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:129 > + int softHyphenPos = bi->string().indexOf(QChar(SOFT_HYPHEN), currPos); this QChar could be static. > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:161 > + // Note that this code will essentially no-op in the case of QTextBoundaryFinder handling soft hyphen. no-op but the indexOf will still be run right? 581857 9 dtharp 2012-03-19 09:22:25 -0700 (In reply to comment #8) > (From update of attachment 132152 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=132152&action=review > > Is the issue reported in Qt? Yes. In fact, the related bug (bug 27593) mentions that this has been fixed in QT (presumably a version greater than the released 4.8 SDK). > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:129 > > + int softHyphenPos = bi->string().indexOf(QChar(SOFT_HYPHEN), currPos); > > this QChar could be static. I was going for compactness since this is somewhat of a corner case. Are you saying I should declare a static instance of a QChar and pass it in instead? > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:161 > > + // Note that this code will essentially no-op in the case of QTextBoundaryFinder handling soft hyphen. > > no-op but the indexOf will still be run right? Yes. By no-op I meant that the if condition would never be met if/when QTextBoundaryFinder handles soft hyphen correctly. 581997 10 menard 2012-03-19 11:43:55 -0700 (In reply to comment #9) > (In reply to comment #8) > > (From update of attachment 132152 [details] [details]) > > View in context: https://bugs.webkit.org/attachment.cgi?id=132152&action=review > > > > Is the issue reported in Qt? > Yes. In fact, the related bug (bug 27593) mentions that this has been fixed in QT (presumably a version greater than the released 4.8 SDK). > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:129 > > > + int softHyphenPos = bi->string().indexOf(QChar(SOFT_HYPHEN), currPos); > > > > this QChar could be static. > I was going for compactness since this is somewhat of a corner case. Are you saying I should declare a static instance of a QChar and pass it in instead? > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:161 > > > + // Note that this code will essentially no-op in the case of QTextBoundaryFinder handling soft hyphen. > > > > no-op but the indexOf will still be run right? > Yes. By no-op I meant that the if condition would never be met if/when QTextBoundaryFinder handles soft hyphen correctly. Provided it is fixed in a patch release of Qt, 4.8.1 I don't think we should put the workaround. Though I'm in favor of adding the test for regression testing. Pierre? 582913 11 pierre.rossi 2012-03-20 03:55:39 -0700 (In reply to comment #10) > (In reply to comment #9) > > (In reply to comment #8) > > > (From update of attachment 132152 [details] [details] [details]) > > > View in context: https://bugs.webkit.org/attachment.cgi?id=132152&action=review > > > > > > Is the issue reported in Qt? > > Yes. In fact, the related bug (bug 27593) mentions that this has been fixed in QT (presumably a version greater than the released 4.8 SDK). > > > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:129 > > > > + int softHyphenPos = bi->string().indexOf(QChar(SOFT_HYPHEN), currPos); > > > > > > this QChar could be static. > > I was going for compactness since this is somewhat of a corner case. Are you saying I should declare a static instance of a QChar and pass it in instead? > > > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:161 > > > > + // Note that this code will essentially no-op in the case of QTextBoundaryFinder handling soft hyphen. > > > > > > no-op but the indexOf will still be run right? > > Yes. By no-op I meant that the if condition would never be met if/when QTextBoundaryFinder handles soft hyphen correctly. > > Provided it is fixed in a patch release of Qt, 4.8.1 I don't think we should put the workaround. Though I'm in favor of adding the test for regression testing. > > Pierre? I had a quick look at the changes that had gone into QTextBoundaryFinder in Qt5 and I'm not sure this has even been fixed. I'm starting to suspect it's only the switch to ICU that made the bug go away for Allan and I. So while I agree with Alexis the test would be nice to have in WebKit, I'm not sure if the proper fix doesn't belong in QTextBoundaryFinder, so that others can benefit from it. 583115 12 dtharp 2012-03-20 09:00:41 -0700 (In reply to comment #11) > (In reply to comment #10) > > (In reply to comment #9) > > > (In reply to comment #8) > > > > (From update of attachment 132152 [details] [details] [details] [details]) > > > > View in context: https://bugs.webkit.org/attachment.cgi?id=132152&action=review > > > > > > > > Is the issue reported in Qt? > > > Yes. In fact, the related bug (bug 27593) mentions that this has been fixed in QT (presumably a version greater than the released 4.8 SDK). > > > > > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:129 > > > > > + int softHyphenPos = bi->string().indexOf(QChar(SOFT_HYPHEN), currPos); > > > > > > > > this QChar could be static. > > > I was going for compactness since this is somewhat of a corner case. Are you saying I should declare a static instance of a QChar and pass it in instead? > > > > > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:161 > > > > > + // Note that this code will essentially no-op in the case of QTextBoundaryFinder handling soft hyphen. > > > > > > > > no-op but the indexOf will still be run right? > > > Yes. By no-op I meant that the if condition would never be met if/when QTextBoundaryFinder handles soft hyphen correctly. > > > > Provided it is fixed in a patch release of Qt, 4.8.1 I don't think we should put the workaround. Though I'm in favor of adding the test for regression testing. > > > > Pierre? > > I had a quick look at the changes that had gone into QTextBoundaryFinder in Qt5 and I'm not sure this has even been fixed. I'm starting to suspect it's only the switch to ICU that made the bug go away for Allan and I. So while I agree with Alexis the test would be nice to have in WebKit, I'm not sure if the proper fix doesn't belong in QTextBoundaryFinder, so that others can benefit from it. I think we are all in heated agreement that this should be fixed in QTextBoundaryFinder :-) I think the fact that it does not appear to be fixed currently in QT presents a good argument for landing this patch: 1. It solves the issue in webkit, in a safe way. 2. The fix is easy to back out later when QT is fixed. 3. The test case is valuable. Is there any reason this patch shouldn't be landed at this point? 585434 13 menard 2012-03-22 10:20:21 -0700 (In reply to comment #12) > (In reply to comment #11) > > (In reply to comment #10) > > > (In reply to comment #9) > > > > (In reply to comment #8) > > > > > (From update of attachment 132152 [details] [details] [details] [details] [details]) > > > > > View in context: https://bugs.webkit.org/attachment.cgi?id=132152&action=review > > > > > > > > > > Is the issue reported in Qt? > > > > Yes. In fact, the related bug (bug 27593) mentions that this has been fixed in QT (presumably a version greater than the released 4.8 SDK). > > > > > > > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:129 > > > > > > + int softHyphenPos = bi->string().indexOf(QChar(SOFT_HYPHEN), currPos); > > > > > > > > > > this QChar could be static. > > > > I was going for compactness since this is somewhat of a corner case. Are you saying I should declare a static instance of a QChar and pass it in instead? > > > > > > > > > > > Source/WebCore/platform/text/qt/TextBreakIteratorQt.cpp:161 > > > > > > + // Note that this code will essentially no-op in the case of QTextBoundaryFinder handling soft hyphen. > > > > > > > > > > no-op but the indexOf will still be run right? > > > > Yes. By no-op I meant that the if condition would never be met if/when QTextBoundaryFinder handles soft hyphen correctly. > > > > > > Provided it is fixed in a patch release of Qt, 4.8.1 I don't think we should put the workaround. Though I'm in favor of adding the test for regression testing. > > > > > > Pierre? > > > > I had a quick look at the changes that had gone into QTextBoundaryFinder in Qt5 and I'm not sure this has even been fixed. I'm starting to suspect it's only the switch to ICU that made the bug go away for Allan and I. So while I agree with Alexis the test would be nice to have in WebKit, I'm not sure if the proper fix doesn't belong in QTextBoundaryFinder, so that others can benefit from it. > > I think we are all in heated agreement that this should be fixed in QTextBoundaryFinder :-) I think the fact that it does not appear to be fixed currently in QT presents a good argument for landing this patch: > 1. It solves the issue in webkit, in a safe way. > 2. The fix is easy to back out later when QT is fixed. > 3. The test case is valuable. > > Is there any reason this patch shouldn't be landed at this point? Yes we add a workaround for a little amount of user. Unreleased QtWebKit version (trunk) building against Qt 4.8.0 which is unsupported/no QA'ed combination. We recommend QtWebKit 2.2 for Qt 4.8.0. Maybe the bug is in QtWebKit 2.2 also but I think it would me more valuable to fix Qt rather than a workaround of the bug in here. At the end it will benefit users of Qt who are not using QtWebKit. You mentioned on IRC that the bug is maybe fix in Qt 4.8 branch then cool no need to add workaround. What you can re-upload is just the test case, that is good to have. 585700 14 dtharp 2012-03-22 14:25:18 -0700 Added bug 81964 for uploading only the test case. Shall we close this bug then? 585702 15 132152 menard 2012-03-22 14:27:04 -0700 Comment on attachment 132152 Patch Clearing review flag as per comments. 623228 16 andreas_nordal_4 2012-05-14 12:48:18 -0700 How to reassign to the Qt people? Isn't this supposed to be the right bugzilla for even Qt specific bugs in QtWebKit? Please advise me. A verified bug like this should stay open _somewhere_ until it gets fixed. I wanted to make sure this was reported to Qt. According to [1], bugs in QtWebKit should be reported to [2], even bugs that are specific to Qt. I could not find it there, but its «Template shortcut» [3] for reporting new bugs redirects to this bugzilla (https://bugs.webkit.org/enter_bug.cgi?…). Also, I found the equivalent of bug 27593 on qt-project.org [4], but the message was clear: «please move this issue to bugs.webkit.org» [1] https://qt-project.org/wiki/ReportingBugsInQt [2] http://trac.webkit.org/wiki/QtWebKitBugs [3] http://webkit.org/new-qtwebkit-bug [4] https://bugreports.qt-project.org/browse/QTWEBKIT-58 624226 17 dtharp 2012-05-15 10:22:02 -0700 (In reply to comment #16) > How to reassign to the Qt people? > > Isn't this supposed to be the right bugzilla for even Qt specific bugs in QtWebKit? Please advise me. A verified bug like this should stay open _somewhere_ until it gets fixed. > > I wanted to make sure this was reported to Qt. According to [1], bugs in QtWebKit should be reported to [2], even bugs that are specific to Qt. I could not find it there, but its «Template shortcut» [3] for reporting new bugs redirects to this bugzilla (https://bugs.webkit.org/enter_bug.cgi?…). > > Also, I found the equivalent of bug 27593 on qt-project.org [4], but the message was clear: «please move this issue to bugs.webkit.org» > > [1] https://qt-project.org/wiki/ReportingBugsInQt > [2] http://trac.webkit.org/wiki/QtWebKitBugs > [3] http://webkit.org/new-qtwebkit-bug > [4] https://bugreports.qt-project.org/browse/QTWEBKIT-58 Just FYI, the patch uploaded for this bug is still valid (fixes the issue in a safe way in webkit -- will be transparent if/when QT fixes it). If there is consensus, I can attempt to get this reviewed again and landed. 626399 18 andreas_nordal_4 2012-05-17 02:36:41 -0700 (In reply to comment #17) > Just FYI, the patch uploaded for this bug is still valid (fixes the issue > in a safe way in webkit -- will be transparent if/when QT fixes it). If > there is consensus, I can attempt to get this reviewed again and landed. If your workaround is harmless, I'm for it, but I'm not competent. As long as the real bug is in QTextBoundaryFinder, we must let the Qt people know. Ideally, someone competent should write a bugreport on the misbehavior of QTextBoundaryFinder. Alternatively, we could hand this report over to them. The latter was what I meant by «reassign to the Qt people». If the problem is reported over at Qt, I'm happy with closing the bug here. Happy 17th May ;) 626660 19 pierre.rossi 2012-05-17 10:00:28 -0700 (In reply to comment #18) > (In reply to comment #17) > > Just FYI, the patch uploaded for this bug is still valid (fixes the issue > > in a safe way in webkit -- will be transparent if/when QT fixes it). If > > there is consensus, I can attempt to get this reviewed again and landed. > > If your workaround is harmless, I'm for it, but I'm not competent. > > As long as the real bug is in QTextBoundaryFinder, we must let the Qt people know. Ideally, someone competent should write a bugreport on the misbehavior of QTextBoundaryFinder. Alternatively, we could hand this report over to them. The latter was what I meant by «reassign to the Qt people». > > If the problem is reported over at Qt, I'm happy with closing the bug here. > > Happy 17th May ;) The problem in QTextBoundaryFinder is now hidden by the fact that we switched to ICU, so this is really not a WebKit bug per se. Here's the fix for Qt5: https://codereview.qt-project.org/#change,26469 And the request for the Qt 4 repo: https://codereview.qt-project.org/#change,26470 As for bug reporting, bugs with the [Qt] keyword are specific to QtWebKit, and people hacking on the Qt port usually also contribute to Qt as a direct consequence. Happy 17th of May indeed :) 747365 20 allan.jensen 2012-10-22 03:05:13 -0700 Any progress on landing the Qt4.8 fix? 747420 21 pierre.rossi 2012-10-22 04:49:07 -0700 (In reply to comment #20) > Any progress on landing the Qt4.8 fix? (In reply to comment #20) > Any progress on landing the Qt4.8 fix? The problem is that it's technically a behavior change, and it's a bit against the policy to land behavior changes in patch releases in Qt, unless we're fixing a critical bug that way. So the fact that we're stuck with 4.8 makes it tricky... Aren't we gonna use ICU by default for QtWebKit 2.3 though ? That should fix it for most people. 747422 22 hausmann 2012-10-22 04:50:48 -0700 (In reply to comment #21) > (In reply to comment #20) > > Any progress on landing the Qt4.8 fix? > > (In reply to comment #20) > > Any progress on landing the Qt4.8 fix? > > The problem is that it's technically a behavior change, and it's a bit against the policy to land behavior changes in patch releases in Qt, unless we're fixing a critical bug that way. So the fact that we're stuck with 4.8 makes it tricky... > Aren't we gonna use ICU by default for QtWebKit 2.3 though ? That should fix it for most people. I agree, I think the dependency to ICU is the best way of solving this. 747437 23 allan.jensen 2012-10-22 05:05:38 -0700 (In reply to comment #22) > (In reply to comment #21) > > (In reply to comment #20) > > > Any progress on landing the Qt4.8 fix? > > > > (In reply to comment #20) > > > Any progress on landing the Qt4.8 fix? > > > > The problem is that it's technically a behavior change, and it's a bit against the policy to land behavior changes in patch releases in Qt, unless we're fixing a critical bug that way. So the fact that we're stuck with 4.8 makes it tricky... > > Aren't we gonna use ICU by default for QtWebKit 2.3 though ? That should fix it for most people. > > I agree, I think the dependency to ICU is the best way of solving this. So far we are not using ICU in QtWebKit 2.3, but it might be a better solution, though I feel sad for not fixing bugs in Qt. 747439 24 pierre.rossi 2012-10-22 05:09:16 -0700 (In reply to comment #23) > (In reply to comment #22) > > (In reply to comment #21) > > > (In reply to comment #20) > > > > Any progress on landing the Qt4.8 fix? > > > > > > (In reply to comment #20) > > > > Any progress on landing the Qt4.8 fix? > > > > > > The problem is that it's technically a behavior change, and it's a bit against the policy to land behavior changes in patch releases in Qt, unless we're fixing a critical bug that way. So the fact that we're stuck with 4.8 makes it tricky... > > > Aren't we gonna use ICU by default for QtWebKit 2.3 though ? That should fix it for most people. > > > > I agree, I think the dependency to ICU is the best way of solving this. > > So far we are not using ICU in QtWebKit 2.3, but it might be a better solution, though I feel sad for not fixing bugs in Qt. Well it's been fixed in Qt5, it's just not something that can be applied to Qt4 nicely at this point. 747440 25 hausmann 2012-10-22 05:10:01 -0700 (In reply to comment #23) > (In reply to comment #22) > > (In reply to comment #21) > > > (In reply to comment #20) > > > > Any progress on landing the Qt4.8 fix? > > > > > > (In reply to comment #20) > > > > Any progress on landing the Qt4.8 fix? > > > > > > The problem is that it's technically a behavior change, and it's a bit against the policy to land behavior changes in patch releases in Qt, unless we're fixing a critical bug that way. So the fact that we're stuck with 4.8 makes it tricky... > > > Aren't we gonna use ICU by default for QtWebKit 2.3 though ? That should fix it for most people. > > > > I agree, I think the dependency to ICU is the best way of solving this. > > So far we are not using ICU in QtWebKit 2.3, but it might be a better solution, though I feel sad for not fixing bugs in Qt. The upside is that QTextBoundaryFinder is getting active feature development and maintenance in Qt 5 from Ritt :) 123786 2012-01-24 12:45:27 -0800 2012-01-24 12:45:27 -0800 testcase demonstrating overlapping text caused by soft hyphen shy.html text/html 339 andreas_nordal_4 PGh0bWw+CjxoZWFkPgo8dGl0bGU+U29mdCBoeXBoZW48L3RpdGxlPgo8L2hlYWQ+Cjxib2R5PgoK PHRhYmxlIGJvcmRlcj0iMSI+Cjx0cj4KPHRkPmFhYWFhYWFhYWFhYWFhJnNoeTtiYmJiYmJiYmJi YmJiYjwvdGQ+Cjx0ZD5zb21ldGhpbmc8L3RkPgo8L3RyPgo8L3RhYmxlPgoKPHA+VXNlIHdlYmtp dC48YnI+CldpbGwgdGhlIGxpbmUgaW4gdGhlIGxlZnQgdGFibGUgY2VsbCBicmVhayBpZiB5b3Ug bmFycm93IHRoZSB3aW5kb3c/PGJyPgpXaWxsIHRoZSAiYmJiYmJiYmJiYmJiYmIiIG1vdmUgb3V0 IG9mIHRoZSB3YXkgZm9yIHRoYXQgInNvbWV0aGluZyI/PC9wPgoKPC9ib2R5Pgo8L2h0bWw+ 132152 2012-03-15 16:57:40 -0700 2012-03-22 14:27:04 -0700 Patch bug-76932-20120315165707.patch text/plain 6629 dtharp U3VidmVyc2lvbiBSZXZpc2lvbjogMTEwOTA4CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggYjYxMGMxMTRhODU1MmI3 YTliY2VmYjI4MDgyMWJhNzdkMzM4ZWM1My4uNGU4NTllOGUyNjIwZThkY2QzNDRmYWJmMWM0NGVk NmNkOTEwZTg3NCAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDIzIEBACisyMDEyLTAzLTE1ICBEYXZl IFRoYXJwICA8ZHRoYXJwQGNvZGVhdXJvcmEub3JnPgorCisgICAgICAgIFtRdF0gc29mdCBoeXBo ZW4gZG9lcyBub3QgYnJlYWsgdGhlIGxpbmUsIGluc3RlYWQgY2F1c2VzIGNvbnRlbnQgdG8gb3Zl cmxhcAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9NzY5 MzIKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBUaGlz IGlzIGEgUVQgZml4IG9ubHkuIEFkZGVkIGNvZGUgdG8gaGFuZGxlIGxpbmUgYnJlYWsgb24gc29m dCBoeXBoZW4gaW4gdGhlIGNhc2UgCisgICAgICAgIHRoYXQgdGhlIFFUIENsYXNzIFFUZXh0Qm91 bmRhcnlGaW5kZXIgZmFpbHMgdG8gcmVzcGVjdCBzb2Z0IGh5cGhlbiAoYXMgaXMgdGhlIGNhc2Ug CisgICAgICAgIGluIFFUIDQuOC4wIGFuZCBwcmV2aW91cykuIE5vdGUgdGhhdCB0aGlzIGNvZGUg d2lsbCBlc3NlbnRpYWxseSBuby1vcCB3aGVuIAorICAgICAgICBRVGV4dEJvdW5kYXJ5RmluZGVy IGJlZ2lucyByZXNwZWN0aW5nIHNvZnQgaHlwaGVuLiAgSSd2ZSBhZGRlZCAnRklYTUUnIGNvbW1l bnRzIGluIAorICAgICAgICB0aGUgY29kZSB0byB0aGlzIGVmZmVjdC4KKworICAgICAgICBUZXN0 OiBmYXN0L3RleHQvc29mdC1oeXBoZW4tNS5odG1sCisKKyAgICAgICAgKiBwbGF0Zm9ybS90ZXh0 L3F0L1RleHRCcmVha0l0ZXJhdG9yUXQuY3BwOgorICAgICAgICAoV2ViQ29yZTo6dGV4dEJyZWFr TmV4dCk6CisgICAgICAgIChXZWJDb3JlOjp0ZXh0QnJlYWtQcmVjZWRpbmcpOgorICAgICAgICAo V2ViQ29yZTo6dGV4dEJyZWFrRm9sbG93aW5nKToKKwogMjAxMi0wMy0xNSAgQW5kZXJzIENhcmxz c29uICA8YW5kZXJzY2FAYXBwbGUuY29tPgogCiAgICAgICAgIG9ubHkgc29tZSB0aWxlcyByZW5k ZXJlZCB3aGVuIHpvb21lZCBpbgpkaWZmIC0tZ2l0IGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0v dGV4dC9xdC9UZXh0QnJlYWtJdGVyYXRvclF0LmNwcCBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3Jt L3RleHQvcXQvVGV4dEJyZWFrSXRlcmF0b3JRdC5jcHAKaW5kZXggNjUzZWRlMzlhYjBhNWE1NGZi NTdiZjA5ODQ5YmY3NDI0ODBjYmMwYi4uMmI1ZjhhMGI1NzQ2OGRkYjhlN2ZhZWUzYzQzOWJiNmZi YTMwODUxZCAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvcGxhdGZvcm0vdGV4dC9xdC9UZXh0 QnJlYWtJdGVyYXRvclF0LmNwcAorKysgYi9Tb3VyY2UvV2ViQ29yZS9wbGF0Zm9ybS90ZXh0L3F0 L1RleHRCcmVha0l0ZXJhdG9yUXQuY3BwCkBAIC0zMiw2ICszMiw4IEBACiAjZGVmaW5lIERFQlVH IGlmICgxKSB7fSBlbHNlIHFEZWJ1ZwogI2VuZGlmCiAKKyNkZWZpbmUgU09GVF9IWVBIRU4gMHhB RAorCiB1c2luZyBuYW1lc3BhY2Ugc3RkOwogCiBuYW1lc3BhY2UgV2ViQ29yZSB7CkBAIC0xMTks NyArMTIxLDE2IEBAIG5hbWVzcGFjZSBXZWJDb3JlIHsKIAogICAgIGludCB0ZXh0QnJlYWtOZXh0 KFRleHRCcmVha0l0ZXJhdG9yKiBiaSkKICAgICB7CisgICAgICAgIGludCBjdXJyUG9zID0gYmkt PnBvc2l0aW9uKCk7CiAgICAgICAgIGludCBwb3MgPSBiaS0+dG9OZXh0Qm91bmRhcnkoKTsKKwor ICAgICAgICAvLyBGSVhNRTogVGhlIGZvbGxvd2luZyBjb2RlIGNhbiBiZSByZW1vdmVkIHdoZW4g UVQncyBRVGV4dEJvdW5kYXJ5RmluZGVyIGNsYXNzIGhhbmRsZXMgc29mdCBoeXBoZW4uCisgICAg ICAgIC8vICAgICAgICBOb3RlIHRoYXQgdGhpcyBjb2RlIHdpbGwgZXNzZW50aWFsbHkgbm8tb3Ag aW4gdGhlIGNhc2Ugb2YgUVRleHRCb3VuZGFyeUZpbmRlciBoYW5kbGluZyBzb2Z0IGh5cGhlbi4K KyAgICAgICAgaW50IHNvZnRIeXBoZW5Qb3MgPSBiaS0+c3RyaW5nKCkuaW5kZXhPZihRQ2hhcihT T0ZUX0hZUEhFTiksIGN1cnJQb3MpOworICAgICAgICBpZiAoYmktPnR5cGUoKSA9PSBRVGV4dEJv dW5kYXJ5RmluZGVyOjpMaW5lICYmIHNvZnRIeXBoZW5Qb3MgIT0gLTEgJiYgY3VyclBvcyA8IHNv ZnRIeXBoZW5Qb3MgJiYgc29mdEh5cGhlblBvcyA8IHBvcykgeworICAgICAgICAgICAgcG9zID0g c29mdEh5cGhlblBvcyArIDE7IC8vIHNraXAgdGhlIHNvZnQtaHlwaGVuCisgICAgICAgICAgICBi aS0+c2V0UG9zaXRpb24ocG9zKTsKKyAgICAgICAgfQogICAgICAgICBERUJVRygpIDw8ICJ0ZXh0 QnJlYWtOZXh0IiA8PCBwb3M7CiAgICAgICAgIHJldHVybiBwb3M7CiAgICAgfQpAQCAtMTI4LDcg KzEzOSwxNiBAQCBuYW1lc3BhY2UgV2ViQ29yZSB7CiAgICAgewogICAgICAgICBiaS0+c2V0UG9z aXRpb24ocG9zKTsKICAgICAgICAgaW50IG5ld3BvcyA9IGJpLT50b1ByZXZpb3VzQm91bmRhcnko KTsKKworICAgICAgICAvLyBGSVhNRTogVGhlIGZvbGxvd2luZyBjb2RlIGNhbiBiZSByZW1vdmVk IHdoZW4gUVQncyBRVGV4dEJvdW5kYXJ5RmluZGVyIGNsYXNzIGhhbmRsZXMgc29mdCBoeXBoZW4u CisgICAgICAgIC8vICAgICAgICBOb3RlIHRoYXQgdGhpcyBjb2RlIHdpbGwgZXNzZW50aWFsbHkg bm8tb3AgaW4gdGhlIGNhc2Ugb2YgUVRleHRCb3VuZGFyeUZpbmRlciBoYW5kbGluZyBzb2Z0IGh5 cGhlbi4KKyAgICAgICAgaW50IHNvZnRIeXBoZW5Qb3MgPSBiaS0+c3RyaW5nKCkuaW5kZXhPZihR Q2hhcihTT0ZUX0hZUEhFTiksIG5ld3Bvcyk7CisgICAgICAgIGlmIChiaS0+dHlwZSgpID09IFFU ZXh0Qm91bmRhcnlGaW5kZXI6OkxpbmUgJiYgc29mdEh5cGhlblBvcyAhPSAtMSAmJiBuZXdwb3Mg PCBzb2Z0SHlwaGVuUG9zICYmIHNvZnRIeXBoZW5Qb3MgPCBwb3MpIHsKKyAgICAgICAgICAgIG5l d3BvcyA9IHNvZnRIeXBoZW5Qb3MgLSAxOyAvLyBza2lwIHRoZSBzb2Z0LWh5cGhlbgorICAgICAg ICAgICAgYmktPnNldFBvc2l0aW9uKG5ld3Bvcyk7CisgICAgICAgIH0KICAgICAgICAgREVCVUco KSA8PCAidGV4dEJyZWFrUHJlY2VkaW5nIiA8PCBwb3MgPDwgbmV3cG9zOworCiAgICAgICAgIHJl dHVybiBuZXdwb3M7CiAgICAgfQogCkBAIC0xMzYsNiArMTU2LDE0IEBAIG5hbWVzcGFjZSBXZWJD b3JlIHsKICAgICB7CiAgICAgICAgIGJpLT5zZXRQb3NpdGlvbihwb3MpOwogICAgICAgICBpbnQg bmV3cG9zID0gYmktPnRvTmV4dEJvdW5kYXJ5KCk7CisKKyAgICAgICAgLy8gRklYTUU6IFRoZSBm b2xsb3dpbmcgY29kZSBjYW4gYmUgcmVtb3ZlZCB3aGVuIFFUJ3MgUVRleHRCb3VuZGFyeUZpbmRl ciBjbGFzcyBoYW5kbGVzIHNvZnQgaHlwaGVuLgorICAgICAgICAvLyAgICAgICAgTm90ZSB0aGF0 IHRoaXMgY29kZSB3aWxsIGVzc2VudGlhbGx5IG5vLW9wIGluIHRoZSBjYXNlIG9mIFFUZXh0Qm91 bmRhcnlGaW5kZXIgaGFuZGxpbmcgc29mdCBoeXBoZW4uCisgICAgICAgIGludCBzb2Z0SHlwaGVu UG9zID0gYmktPnN0cmluZygpLmluZGV4T2YoUUNoYXIoU09GVF9IWVBIRU4pLCBwb3MpOworICAg ICAgICBpZiAoYmktPnR5cGUoKSA9PSBRVGV4dEJvdW5kYXJ5RmluZGVyOjpMaW5lICYmIHNvZnRI eXBoZW5Qb3MgIT0gLTEgJiYgcG9zIDwgc29mdEh5cGhlblBvcyAmJiBzb2Z0SHlwaGVuUG9zIDwg bmV3cG9zKSB7CisgICAgICAgICAgICBuZXdwb3MgPSBzb2Z0SHlwaGVuUG9zICsgMTsgLy8gc2tp cCB0aGUgc29mdC1oeXBoZW4KKyAgICAgICAgICAgIGJpLT5zZXRQb3NpdGlvbihuZXdwb3MpOwor ICAgICAgICB9CiAgICAgICAgIERFQlVHKCkgPDwgInRleHRCcmVha0ZvbGxvd2luZyIgPDwgcG9z IDw8IG5ld3BvczsKICAgICAgICAgcmV0dXJuIG5ld3BvczsKICAgICB9CmRpZmYgLS1naXQgYS9M YXlvdXRUZXN0cy9DaGFuZ2VMb2cgYi9MYXlvdXRUZXN0cy9DaGFuZ2VMb2cKaW5kZXggNzlhMjU2 MDNiZmNkZDllNDgyZjU1YmE3OTdlNDBkMjBjMTA3ZDZhYi4uMzBhYjNkNTYyMjk2NjM3ZWMzNTgz ZjNhODQzODdiMTY3YjdhMmNmNCAxMDA2NDQKLS0tIGEvTGF5b3V0VGVzdHMvQ2hhbmdlTG9nCisr KyBiL0xheW91dFRlc3RzL0NoYW5nZUxvZwpAQCAtMSwzICsxLDE1IEBACisyMDEyLTAzLTE1ICBE YXZlIFRoYXJwICA8ZHRoYXJwQGNvZGVhdXJvcmEub3JnPgorCisgICAgICAgIFtRdF0gc29mdCBo eXBoZW4gZG9lcyBub3QgYnJlYWsgdGhlIGxpbmUsIGluc3RlYWQgY2F1c2VzIGNvbnRlbnQgdG8g b3ZlcmxhcAorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9 NzY5MzIKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBB ZGRlZCBhIG5ldyBzb2Z0IGh5cGhlbiB0ZXN0IHRvIGNhdGNoIHRoZSBzb2Z0IGh5cGhlbiBpc3N1 ZSBmb3VuZCBvbiBRVC4KKworICAgICAgICAqIGZhc3QvdGV4dC9zb2Z0LWh5cGhlbi01LWV4cGVj dGVkLnR4dDogQWRkZWQuCisgICAgICAgICogZmFzdC90ZXh0L3NvZnQtaHlwaGVuLTUuaHRtbDog QWRkZWQuCisKIDIwMTItMDMtMTUgIEplc3NpZSBCZXJsaW4gIDxqYmVybGluQGFwcGxlLmNvbT4K IAogICAgICAgICBTZWxlY3Rpb24gaXMgbm90IGNvbGxhcHNlZCBpbiBzb21lIFdLMiBlZGl0aW5n IHRlc3RzCmRpZmYgLS1naXQgYS9MYXlvdXRUZXN0cy9mYXN0L3RleHQvc29mdC1oeXBoZW4tNS1l eHBlY3RlZC50eHQgYi9MYXlvdXRUZXN0cy9mYXN0L3RleHQvc29mdC1oeXBoZW4tNS1leHBlY3Rl ZC50eHQKbmV3IGZpbGUgbW9kZSAxMDA2NDQKaW5kZXggMDAwMDAwMDAwMDAwMDAwMDAwMDAwMDAw MDAwMDAwMDAwMDAwMDAwMC4uOTQzN2RjYjk4NDNjYTZiY2RhZDZmODEyMjE2ZWE0NGU5ZDRiMTQ0 NQotLS0gL2Rldi9udWxsCisrKyBiL0xheW91dFRlc3RzL2Zhc3QvdGV4dC9zb2Z0LWh5cGhlbi01 LWV4cGVjdGVkLnR4dApAQCAtMCwwICsxLDEwIEBACitTb2Z0IEh5cGhlbiBUZXN0CisKK09uIHN1 Y2Nlc3MsIHlvdSB3aWxsIHNlZSBhIHNlcmllcyBvZiAiUEFTUyIgbWVzc2FnZXMsIGZvbGxvd2Vk IGJ5ICJURVNUIENPTVBMRVRFIi4KKworCitQQVNTIHN1Y2Nlc3NmdWxseVBhcnNlZCBpcyB0cnVl CisKK1RFU1QgQ09NUExFVEUKK1BBU1MgdHJ1ZSBpcyB0cnVlCithbnRpwq1kaXPCrWVzdMKtYWLC rWxpc2jCrW1lbnTCrWFyaWFuwq1pc20uCmRpZmYgLS1naXQgYS9MYXlvdXRUZXN0cy9mYXN0L3Rl eHQvc29mdC1oeXBoZW4tNS5odG1sIGIvTGF5b3V0VGVzdHMvZmFzdC90ZXh0L3NvZnQtaHlwaGVu LTUuaHRtbApuZXcgZmlsZSBtb2RlIDEwMDY0NAppbmRleCAwMDAwMDAwMDAwMDAwMDAwMDAwMDAw MDAwMDAwMDAwMDAwMDAwMDAwLi5mYjU5ZjdkNDE3ODU4YjE1ZGRhMmM0OGVlMjkwN2QxNTRiMDM1 NzlmCi0tLSAvZGV2L251bGwKKysrIGIvTGF5b3V0VGVzdHMvZmFzdC90ZXh0L3NvZnQtaHlwaGVu LTUuaHRtbApAQCAtMCwwICsxLDI5IEBACis8IURPQ1RZUEUgaHRtbD4KKzxodG1sPgorPGhlYWQ+ Cis8bWV0YSBjaGFyc2V0PSJ1dGYtOCI+Cis8c2NyaXB0IHNyYz0iLi4vanMvcmVzb3VyY2VzL2pz LXRlc3QtcHJlLmpzIj48L3NjcmlwdD4KKzwvaGVhZD4KKzxib2R5IG9ubG9hZD0icnVuVGVzdCgp OyI+CisgICAgPHNjcmlwdD4KKyAgICAgICAgZGVzY3JpcHRpb24oIlNvZnQgSHlwaGVuIFRlc3Qi KTsKKworICAgICAgICBmdW5jdGlvbiBydW5UZXN0KCkgeworICAgICAgICAgICAgaWYgKHdpbmRv dy5sYXlvdXRUZXN0Q29udHJvbGxlcikKKyAgICAgICAgICAgICAgICBsYXlvdXRUZXN0Q29udHJv bGxlci5kdW1wQXNUZXh0KCk7CisKKyAgICAgICAgICAgIC8vIDU2IGlzIDQgKiB0aGUgZm9udCBz aXplICgxNCkuIFdlJ3JlIGV4cGVjdGluZyB0aGUgdGV4dCB0byBicmVhayB1cCBpbnRvIDQgbGlu ZXMuCisgICAgICAgICAgICAvLyA0IGxpbmVzICsgdGhlIHBhZGRpbmcgZ2l2ZXMgYSBkaXYgaGVp Z2h0IG9mIDkyIGluIHdlYmtpdCBhbmQgRkYsIDg4IGluIGNocm9tZS4KKyAgICAgICAgICAgIC8v IFNvIGFueXRoaW5nIGxlc3MgdGhhbiA1NiBpcyBhIEZBSUwuCisgICAgICAgICAgICB2YXIgZG9l c0l0UGFzcyA9IEJvb2xlYW4oZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoJ3RleHQxJykub2Zmc2V0 SGVpZ2h0ID4gNTYpIDsKKyAgICAgICAgICAgIHNob3VsZEJlVHJ1ZShTdHJpbmcoZG9lc0l0UGFz cykpIDsKKyAgICAgICAgfQorICAgIDwvc2NyaXB0PgorCisgICAgPGRpdiBpZD0idGV4dDEiIHN0 eWxlPSJ3aWR0aDoxNTBweDsgZm9udC1mYW1pbHk6QWhlbTsgZm9udC1zaXplOjE0cHg7IGJvcmRl cjoycHggc29saWQgcmVkIj4KKyAgICAgICAgPHA+YW50aSZzaHk7ZGlzJnNoeTtlc3Qmc2h5O2Fi JnNoeTtsaXNoJnNoeTttZW50JnNoeTthcmlhbiZzaHk7aXNtLjwvcD4KKyAgICA8L2Rpdj4KKwor PHNjcmlwdCBzcmM9Ii4uL2pzL3Jlc291cmNlcy9qcy10ZXN0LXBvc3QuanMiPjwvc2NyaXB0Pgor PC9ib2R5PgorPC9odG1sPgo=