49040 2010-11-04 17:04:11 -0700 Pasting large amounts of plain text in a text area is very slow 2010-11-05 17:15:23 -0700 1 1 1 Unclassified WebKit HTML Editing 528+ (Nightly build) All OS X 10.5 RESOLVED FIXED InRadar P2 Normal --- 1 enrica enrica adele ojan rniwa tony oldest_to_newest 305014 0 enrica 2010-11-04 17:04:11 -0700 Select a large amount of plain text (~500k) from a file and copy Open the attached file. Paste into the text area. Result: the browser hangs for several seconds. 305015 1 73006 enrica 2010-11-04 17:04:40 -0700 Created attachment 73006 Test 305017 2 73007 enrica 2010-11-04 17:11:26 -0700 Created attachment 73007 Patch 305021 3 73007 adele 2010-11-04 17:20:51 -0700 Comment on attachment 73007 Patch Would we also want this optimization if we're pasting into any element that has user-modify: read-write-plaintext-only set? 305161 4 ap 2010-11-04 23:21:41 -0700 See also: bug 34237 (a case of slow pasting that Enrica fixed a few months ago). 305164 5 73007 rniwa 2010-11-04 23:34:35 -0700 Comment on attachment 73007 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=73007&action=review > WebCore/ChangeLog:11 > + pasting into a text area. What we really avoid is iterating through > + a large number of paragraph when no style change is needed. "What we really avoid is iterating through..." sounds confusing to me. Maybe just "We avoid iterating through..."? > WebCore/editing/ReplaceSelectionCommand.cpp:1164 > - if (m_matchStyle) { > + if (m_matchStyle && !isNodeInTextFormControl(start.node())) { Isn't better to set m_matchStyle=false right at the front if we're in a text form control? Is there a case we need m_matchStyle=true even when we're in a text form control? 305335 6 tony 2010-11-05 10:03:58 -0700 It might be possible to write a perf test for this (seems like the time should be constant or linear). See LayoutTests/resources/magnitude-perf.js and LayoutTests/perf. 305347 7 enrica 2010-11-05 10:23:43 -0700 (In reply to comment #3) > (From update of attachment 73007 [details]) > Would we also want this optimization if we're pasting into any element that has user-modify: read-write-plaintext-only set? I'll look into it. 305351 8 enrica 2010-11-05 10:27:59 -0700 (In reply to comment #5) > (From update of attachment 73007 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=73007&action=review > > > WebCore/ChangeLog:11 > > + pasting into a text area. What we really avoid is iterating through > > + a large number of paragraph when no style change is needed. > > "What we really avoid is iterating through..." sounds confusing to me. Maybe just "We avoid iterating through..."? > Not sure that what you suggest is less confusing. I can rephrase this. The point I'm trying to make is that the performance hit comes from the paragraph iteration. > > WebCore/editing/ReplaceSelectionCommand.cpp:1164 > > - if (m_matchStyle) { > > + if (m_matchStyle && !isNodeInTextFormControl(start.node())) { > > Isn't better to set m_matchStyle=false right at the front if we're in a text form control? I believe it makes no difference, this is where we apply the style, if we check here we skip applying the style. Is there a case we need m_matchStyle=true even when we're in a text form control? No. In general we wouldn't need to apply the style when pasting plain text except for the case where we want to preserve the typing style. 305353 9 enrica 2010-11-05 10:30:41 -0700 (In reply to comment #6) > It might be possible to write a perf test for this (seems like the time should be constant or linear). See LayoutTests/resources/magnitude-perf.js and LayoutTests/perf. This fix doesn't change the performance from non-linear to linear. The performance is linear already. Do you think this is still worth? 305374 10 rniwa 2010-11-05 10:51:15 -0700 (In reply to comment #8) > No. In general we wouldn't need to apply the style when pasting plain text except for the case where we want to preserve the typing style. Ok, in that case, it's cleaner to set m_matchStyle false if we're in a text form control. When we don't need this extra condition in the if statement. I think setting this either when we instantiates ReplaceSelectionCommand or right at the beginning of doApply will also prevent us from needlessly obtaining the computed style. 305377 11 tony 2010-11-05 10:53:51 -0700 (In reply to comment #9) > (In reply to comment #6) > > It might be possible to write a perf test for this (seems like the time should be constant or linear). See LayoutTests/resources/magnitude-perf.js and LayoutTests/perf. > > This fix doesn't change the performance from non-linear to linear. The performance is linear already. > Do you think this is still worth? It might be fast enough to show up as constant even though it's technically linear. Another way to think of it is testing to make sure that pasting 500k of text always takes less than X ms. I'm not sure that magnitude-perf.js will handle this type of test well, Ojan, what do you think? 305385 12 rniwa 2010-11-05 11:00:58 -0700 (In reply to comment #11) > It might be fast enough to show up as constant even though it's technically linear. Another way to think of it is testing to make sure that pasting 500k of text always takes less than X ms. I'm afraid setting a hard time limit will make this test flaky. I'm okay with not having a test for this change. 305479 13 ojan 2010-11-05 13:19:22 -0700 (In reply to comment #11) > (In reply to comment #9) > > (In reply to comment #6) > > > It might be possible to write a perf test for this (seems like the time should be constant or linear). See LayoutTests/resources/magnitude-perf.js and LayoutTests/perf. > > > > This fix doesn't change the performance from non-linear to linear. The performance is linear already. > > Do you think this is still worth? > > It might be fast enough to show up as constant even though it's technically linear. Another way to think of it is testing to make sure that pasting 500k of text always takes less than X ms. > > I'm not sure that magnitude-perf.js will handle this type of test well, Ojan, what do you think? The magnitude perf stuff is still pretty experimental (e.g. these tests are skipped on non-chromium bots right now). I welcome people using it and writing new tests, but it's still on my todo list to make the tests more reliable. Given that this patch is not changing the order of magnitude, I probably wouldn't bother in this case. (In reply to comment #12) > (In reply to comment #11) > > It might be fast enough to show up as constant even though it's technically linear. Another way to think of it is testing to make sure that pasting 500k of text always takes less than X ms. > > I'm afraid setting a hard time limit will make this test flaky. I'm okay with not having a test for this change. We could at least have a manual test. It's not ideal, but it gives a comparison point that can be used to identify regressions between nightly builds. Eventually, if we have a performance testing framework, we can move this test into it. 305494 14 73111 enrica 2010-11-05 13:42:27 -0700 Created attachment 73111 Patch2 This is a new patch that addresses Adele's and Ryosuke's comments. I don't think we need another layout test for this fix, since there are several tests that verify the correct behavior in this case. 305527 15 73111 rniwa 2010-11-05 14:26:45 -0700 Comment on attachment 73111 Patch2 View in context: https://bugs.webkit.org/attachment.cgi?id=73111&action=review > editing/ReplaceSelectionCommand.cpp:804 > + if (isNodeInTextFormControl(selection.start().node()) || (selection.start().node()->renderer() && selection.start().node()->renderer()->style()->userModify() == READ_WRITE_PLAINTEXT_ONLY)) > + m_matchStyle = false; Maybe we should also check the selection end? Or is there guarantee that this would never happen? 305531 16 73111 adele 2010-11-05 14:31:25 -0700 Comment on attachment 73111 Patch2 View in context: https://bugs.webkit.org/attachment.cgi?id=73111&action=review >> editing/ReplaceSelectionCommand.cpp:804 >> + m_matchStyle = false; > > Maybe we should also check the selection end? Or is there guarantee that this would never happen? Is there a need to still do the isNodeInTextFormControl check? 305636 17 73142 enrica 2010-11-05 16:51:49 -0700 Created attachment 73142 Patch3 All tests pass. 305639 18 73142 adele 2010-11-05 16:55:29 -0700 Comment on attachment 73142 Patch3 View in context: https://bugs.webkit.org/attachment.cgi?id=73142&action=review > ChangeLog:19 > + and force no style matching. You should update this comment. 305640 19 rniwa 2010-11-05 16:56:10 -0700 (In reply to comment #17) > Created an attachment (id=73142) [details] > Patch3 > > All tests pass. Your patch is made in WebCore/. Could you remake it in /? 305641 20 73142 rniwa 2010-11-05 16:58:40 -0700 Comment on attachment 73142 Patch3 View in context: https://bugs.webkit.org/attachment.cgi?id=73142&action=review > editing/ReplaceSelectionCommand.cpp:805 > + if ((selection.start().node()->renderer() && selection.start().node()->renderer()->style()->userModify() == READ_WRITE_PLAINTEXT_ONLY) && > + (selection.end().node()->renderer() && selection.end().node()->renderer()->style()->userModify() == READ_WRITE_PLAINTEXT_ONLY)) > + m_matchStyle = false; I'm just curious as to why this is superior to checking that it's in the text form control. Could you elaborate? Of course, this patch LGTM nonetheless. 305644 21 adele 2010-11-05 17:00:19 -0700 Its superior because this optimization will apply to other plaintext editable areas besides form controls. (In reply to comment #20) > (From update of attachment 73142 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=73142&action=review > > > editing/ReplaceSelectionCommand.cpp:805 > > + if ((selection.start().node()->renderer() && selection.start().node()->renderer()->style()->userModify() == READ_WRITE_PLAINTEXT_ONLY) && > > + (selection.end().node()->renderer() && selection.end().node()->renderer()->style()->userModify() == READ_WRITE_PLAINTEXT_ONLY)) > > + m_matchStyle = false; > > I'm just curious as to why this is superior to checking that it's in the text form control. Could you elaborate? Of course, this patch LGTM nonetheless. 305648 22 rniwa 2010-11-05 17:03:52 -0700 (In reply to comment #21) > Its superior because this optimization will apply to other plaintext editable areas besides form controls. Right, but what are they? What uses editable region other than contenteditable and forms? 305650 23 adele 2010-11-05 17:08:27 -0700 Any element could have this style. Just like any element could set "-webkit-user-modify: read-write" to get rich text editing. (In reply to comment #22) > (In reply to comment #21) > > Its superior because this optimization will apply to other plaintext editable areas besides form controls. > > Right, but what are they? What uses editable region other than contenteditable and forms? 305651 24 rniwa 2010-11-05 17:09:56 -0700 (In reply to comment #23) > Any element could have this style. Just like any element could set "-webkit-user-modify: read-write" to get rich text editing. That makes sense. Thanks for the information. 305655 25 enrica 2010-11-05 17:15:23 -0700 Committed revision 71459. 73006 2010-11-04 17:04:40 -0700 2010-11-04 17:04:40 -0700 Test textarea.html text/html 48 enrica UGFzdGUgaGVyZToKPHRleHRhcmVhIGlkPSJ0ZXh0YXJlYSI+PC90ZXh0YXJlYT4K 73007 2010-11-04 17:11:26 -0700 2010-11-05 13:42:27 -0700 Patch fastpaste.txt text/plain 1652 enrica SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA3MTM2NikKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMjEgQEAKKzIwMTAtMTEtMDQgIEVucmljYSBDYXN1Y2NpICA8ZW5yaWNhQGFwcGxl LmNvbT4KKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICBQ YXN0aW5nIGxhcmdlIGFtb3VudHMgb2YgcGxhaW4gdGV4dCBpbiBhIHRleHQgYXJlYSBpcyB2ZXJ5 IHNsb3cKKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTQ5 MDQwCisgICAgICAgIDxyZGFyOi8vcHJvYmxlbS83NjQwOTEyPgorICAgICAgICAKKyAgICAgICAg VGhlIG9wdGltaXphdGlvbiBjb25zaXN0cyBpbiBza2lwcGluZyB0aGUgQXBwbHlTdHlsZSBjb21t YW5kIHdoZW4KKyAgICAgICAgcGFzdGluZyBpbnRvIGEgdGV4dCBhcmVhLiBXaGF0IHdlIHJlYWxs eSBhdm9pZCBpcyBpdGVyYXRpbmcgdGhyb3VnaAorICAgICAgICBhIGxhcmdlIG51bWJlciBvZiBw YXJhZ3JhcGggd2hlbiBubyBzdHlsZSBjaGFuZ2UgaXMgbmVlZGVkLgorICAgICAgICBBIGZ1cnRo ZXIgc3RlcCB3aWxsIGJlIHRvIGltcHJvdmUgdGhlIHBhcmFncmFwaCBpdGVyYXRpb24gd2hlbiBp dAorICAgICAgICBjYW4ndCBiZSBhdm9pZGVkLgorICAgICAgICAKKyAgICAgICAgKiBlZGl0aW5n L1JlcGxhY2VTZWxlY3Rpb25Db21tYW5kLmNwcDoKKyAgICAgICAgKFdlYkNvcmU6OlJlcGxhY2VT ZWxlY3Rpb25Db21tYW5kOjpjb21wbGV0ZUhUTUxSZXBsYWNlbWVudCk6IENoZWNrIGlmIHRoZSBz ZWxlY3Rpb24KKyAgICAgICAgaXMgaW4gYSB0ZXh0IGZvcm0gY29udHJvbC4KKwogMjAxMC0xMS0w NCAgS2F2aXRhIEthbmV0a2FyICA8a2thbmV0a2FyQGNocm9taXVtLm9yZz4KIAogICAgICAgICBS ZXZpZXdlZCBieSBBZGFtIEJhcnRoLgpJbmRleDogV2ViQ29yZS9lZGl0aW5nL1JlcGxhY2VTZWxl Y3Rpb25Db21tYW5kLmNwcAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBXZWJDb3JlL2VkaXRpbmcvUmVwbGFjZVNl bGVjdGlvbkNvbW1hbmQuY3BwCShyZXZpc2lvbiA3MDc4NikKKysrIFdlYkNvcmUvZWRpdGluZy9S ZXBsYWNlU2VsZWN0aW9uQ29tbWFuZC5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTExNjEsNyArMTE2 MSw3IEBAIHZvaWQgUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQ6OmNvbXBsZXRlSFQKICAgICAgICAg cmViYWxhbmNlV2hpdGVzcGFjZUF0KHN0YXJ0KTsKICAgICAgICAgcmViYWxhbmNlV2hpdGVzcGFj ZUF0KGVuZCk7CiAKLSAgICAgICAgaWYgKG1fbWF0Y2hTdHlsZSkgeworICAgICAgICBpZiAobV9t YXRjaFN0eWxlICYmICFpc05vZGVJblRleHRGb3JtQ29udHJvbChzdGFydC5ub2RlKCkpKSB7CiAg ICAgICAgICAgICBBU1NFUlQobV9pbnNlcnRpb25TdHlsZSk7CiAgICAgICAgICAgICBhcHBseVN0 eWxlKG1faW5zZXJ0aW9uU3R5bGUuZ2V0KCksIHN0YXJ0LCBlbmQpOwogICAgICAgICB9ICAgIAo= 73111 2010-11-05 13:42:27 -0700 2010-11-05 16:51:49 -0700 Patch2 fastpaste.txt text/plain 2009 enrica SW5kZXg6IGVkaXRpbmcvUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwCj09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0t IGVkaXRpbmcvUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwCShyZXZpc2lvbiA3MTQ0MCkKKysr IGVkaXRpbmcvUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC03 OTksNiArNzk5LDEwIEBACiAgICAgaWYgKHBlcmZvcm1Ucml2aWFsUmVwbGFjZShmcmFnbWVudCkp CiAgICAgICAgIHJldHVybjsKICAgICAKKyAgICAvLyBXZSBjYW4gc2tpcCBtYXRjaGluZyB0aGUg c3R5bGUgaWYgdGhlIHNlbGVjdGlvbiBpcyBpbiBhIHRleHQgY29udHJvbCBvciBoYXMgcmVhZC13 cml0ZS1wbGFpbnRleHQtb25seSBzdHlsZS4KKyAgICBpZiAoaXNOb2RlSW5UZXh0Rm9ybUNvbnRy b2woc2VsZWN0aW9uLnN0YXJ0KCkubm9kZSgpKSB8fCAoc2VsZWN0aW9uLnN0YXJ0KCkubm9kZSgp LT5yZW5kZXJlcigpICYmIHNlbGVjdGlvbi5zdGFydCgpLm5vZGUoKS0+cmVuZGVyZXIoKS0+c3R5 bGUoKS0+dXNlck1vZGlmeSgpID09IFJFQURfV1JJVEVfUExBSU5URVhUX09OTFkpKQorICAgICAg ICBtX21hdGNoU3R5bGUgPSBmYWxzZTsKKyAgICAKICAgICBpZiAobV9tYXRjaFN0eWxlKQogICAg ICAgICBtX2luc2VydGlvblN0eWxlID0gQXBwbHlTdHlsZUNvbW1hbmQ6OmVkaXRpbmdTdHlsZUF0 UG9zaXRpb24oc2VsZWN0aW9uLnN0YXJ0KCksIEluY2x1ZGVUeXBpbmdTdHlsZSk7CiAgICAgCklu ZGV4OiBDaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gQ2hhbmdlTG9nCShyZXZpc2lvbiA3MTQ0MCkK KysrIENoYW5nZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDIzIEBACisyMDEwLTExLTA0 ICBFbnJpY2EgQ2FzdWNjaSAgPGVucmljYUBhcHBsZS5jb20+CisKKyAgICAgICAgUmV2aWV3ZWQg YnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAgICAgUGFzdGluZyBsYXJnZSBhbW91bnRzIG9mIHBs YWluIHRleHQgaW4gYSB0ZXh0IGFyZWEgaXMgdmVyeSBzbG93CisgICAgICAgIGh0dHBzOi8vYnVn cy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9pZD00OTA0MAorICAgICAgICA8cmRhcjovL3Byb2Js ZW0vNzY0MDkxMj4KKyAgICAgICAgCisgICAgICAgIFRoZSBvcHRpbWl6YXRpb24gY29uc2lzdHMg aW4gc2tpcHBpbmcgdGhlIEFwcGx5U3R5bGUgY29tbWFuZCB3aGVuCisgICAgICAgIHBhc3Rpbmcg aW50byBhIHRleHQgYXJlYSBvciBpbnRvIGFuIGVsZW1lbnQgdGhhdCBoYXMgLXdlYmtpdC11c2Vy LW1vZGlmeTogcmVhZC13cml0ZS1wbGFpbnRleHQtb25seS4KKyAgICAgICAgVGhlIGJpZ2dlc3Qg cGVyZm9ybWFuY2UgaGl0IGNvbWVzIGZyb20gaXRlcmF0aW5nIHRocm91Z2gKKyAgICAgICAgYSBs YXJnZSBudW1iZXIgb2YgcGFyYWdyYXBoIHdoZW4gbm8gc3R5bGUgY2hhbmdlIGlzIG5lZWRlZC4K KyAgICAgICAgQSBmdXJ0aGVyIHN0ZXAgd2lsbCBiZSB0byBpbXByb3ZlIHRoZSBwYXJhZ3JhcGgg aXRlcmF0aW9uIHdoZW4gaXQKKyAgICAgICAgY2FuJ3QgYmUgYXZvaWRlZC4KKyAgICAgICAgCisg ICAgICAgICogZWRpdGluZy9SZXBsYWNlU2VsZWN0aW9uQ29tbWFuZC5jcHA6CisgICAgICAgIChX ZWJDb3JlOjpSZXBsYWNlU2VsZWN0aW9uQ29tbWFuZDo6ZG9BcHBseSk6IENoZWNrIGlmIHRoZSBz ZWxlY3Rpb24KKyAgICAgICAgaXMgaW4gYSB0ZXh0IGZvcm0gY29udHJvbCBvciBpbiBhbiBlbGVt ZW50IHRoYXQgaGFzIC13ZWJraXQtdXNlci1tb2RpZnk6IHJlYWQtd3JpdGUtcGxhaW50ZXh0LW9u bHkKKyAgICAgICAgYW5kIGZvcmNlIG5vIHN0eWxlIG1hdGNoaW5nLgorCiAyMDEwLTExLTA1ICBD aGFybGllIFJlaXMgIDxjcmVpc0BjaHJvbWl1bS5vcmc+CiAKICAgICAgICAgUmV2aWV3ZWQgYnkg RGFyaW4gRmlzaGVyLgo= 73142 2010-11-05 16:51:49 -0700 2010-11-05 16:58:40 -0700 Patch3 fastpaste.txt text/plain 2049 enrica SW5kZXg6IGVkaXRpbmcvUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwCj09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0t IGVkaXRpbmcvUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwCShyZXZpc2lvbiA3MTQ0MCkKKysr IGVkaXRpbmcvUmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwCSh3b3JraW5nIGNvcHkpCkBAIC03 OTksNiArNzk5LDExIEBACiAgICAgaWYgKHBlcmZvcm1Ucml2aWFsUmVwbGFjZShmcmFnbWVudCkp CiAgICAgICAgIHJldHVybjsKICAgICAKKyAgICAvLyBXZSBjYW4gc2tpcCBtYXRjaGluZyB0aGUg c3R5bGUgaWYgdGhlIHNlbGVjdGlvbiBpcyBwbGFpbiB0ZXh0LgorICAgIGlmICgoc2VsZWN0aW9u LnN0YXJ0KCkubm9kZSgpLT5yZW5kZXJlcigpICYmIHNlbGVjdGlvbi5zdGFydCgpLm5vZGUoKS0+ cmVuZGVyZXIoKS0+c3R5bGUoKS0+dXNlck1vZGlmeSgpID09IFJFQURfV1JJVEVfUExBSU5URVhU X09OTFkpICYmCisgICAgICAgIChzZWxlY3Rpb24uZW5kKCkubm9kZSgpLT5yZW5kZXJlcigpICYm IHNlbGVjdGlvbi5lbmQoKS5ub2RlKCktPnJlbmRlcmVyKCktPnN0eWxlKCktPnVzZXJNb2RpZnko KSA9PSBSRUFEX1dSSVRFX1BMQUlOVEVYVF9PTkxZKSkKKyAgICAgICAgbV9tYXRjaFN0eWxlID0g ZmFsc2U7CisgICAgCiAgICAgaWYgKG1fbWF0Y2hTdHlsZSkKICAgICAgICAgbV9pbnNlcnRpb25T dHlsZSA9IEFwcGx5U3R5bGVDb21tYW5kOjplZGl0aW5nU3R5bGVBdFBvc2l0aW9uKHNlbGVjdGlv bi5zdGFydCgpLCBJbmNsdWRlVHlwaW5nU3R5bGUpOwogICAgIApJbmRleDogQ2hhbmdlTG9nCj09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT0KLS0tIENoYW5nZUxvZwkocmV2aXNpb24gNzE0NDApCisrKyBDaGFuZ2VMb2cJKHdv cmtpbmcgY29weSkKQEAgLTEsMyArMSwyMyBAQAorMjAxMC0xMS0wNCAgRW5yaWNhIENhc3VjY2kg IDxlbnJpY2FAYXBwbGUuY29tPgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAoT09QUyEp LgorCisgICAgICAgIFBhc3RpbmcgbGFyZ2UgYW1vdW50cyBvZiBwbGFpbiB0ZXh0IGluIGEgdGV4 dCBhcmVhIGlzIHZlcnkgc2xvdworICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93 X2J1Zy5jZ2k/aWQ9NDkwNDAKKyAgICAgICAgPHJkYXI6Ly9wcm9ibGVtLzc2NDA5MTI+CisgICAg ICAgIAorICAgICAgICBUaGUgb3B0aW1pemF0aW9uIGNvbnNpc3RzIGluIHNraXBwaW5nIHRoZSBB cHBseVN0eWxlIGNvbW1hbmQgd2hlbgorICAgICAgICBwYXN0aW5nIGludG8gYSB0ZXh0IGFyZWEg b3IgaW50byBhbiBlbGVtZW50IHRoYXQgaGFzIC13ZWJraXQtdXNlci1tb2RpZnk6IHJlYWQtd3Jp dGUtcGxhaW50ZXh0LW9ubHkuCisgICAgICAgIFRoZSBiaWdnZXN0IHBlcmZvcm1hbmNlIGhpdCBj b21lcyBmcm9tIGl0ZXJhdGluZyB0aHJvdWdoCisgICAgICAgIGEgbGFyZ2UgbnVtYmVyIG9mIHBh cmFncmFwaCB3aGVuIG5vIHN0eWxlIGNoYW5nZSBpcyBuZWVkZWQuCisgICAgICAgIEEgZnVydGhl ciBzdGVwIHdpbGwgYmUgdG8gaW1wcm92ZSB0aGUgcGFyYWdyYXBoIGl0ZXJhdGlvbiB3aGVuIGl0 CisgICAgICAgIGNhbid0IGJlIGF2b2lkZWQuCisgICAgICAgIAorICAgICAgICAqIGVkaXRpbmcv UmVwbGFjZVNlbGVjdGlvbkNvbW1hbmQuY3BwOgorICAgICAgICAoV2ViQ29yZTo6UmVwbGFjZVNl bGVjdGlvbkNvbW1hbmQ6OmRvQXBwbHkpOiBDaGVjayBpZiB0aGUgc2VsZWN0aW9uCisgICAgICAg IGlzIGluIGEgdGV4dCBmb3JtIGNvbnRyb2wgb3IgaW4gYW4gZWxlbWVudCB0aGF0IGhhcyAtd2Vi a2l0LXVzZXItbW9kaWZ5OiByZWFkLXdyaXRlLXBsYWludGV4dC1vbmx5CisgICAgICAgIGFuZCBm b3JjZSBubyBzdHlsZSBtYXRjaGluZy4KKwogMjAxMC0xMS0wNSAgQ2hhcmxpZSBSZWlzICA8Y3Jl aXNAY2hyb21pdW0ub3JnPgogCiAgICAgICAgIFJldmlld2VkIGJ5IERhcmluIEZpc2hlci4K