52044 2011-01-06 21:15:01 -0800 REGRESSION(r74807): No-store is ignored within a document 2011-01-10 10:45:16 -0800 1 1 1 Unclassified WebKit Page Loading 528+ (Nightly build) All All RESOLVED FIXED Regression P1 Normal --- 1 ap ap darin koivisto oldest_to_newest 330641 0 ap 2011-01-06 21:15:01 -0800 Starting with r74807 ("Move loading related code from MemoryCache to CachedResourceLoader"), requests for the same resource URL always return cached data, even if the resource is served as no-store. This doesn't match Firefox, and is just wrong. I expect this to break sites, even though I do not have any examples (this change isn't even in nightlies yet). 330642 1 78207 ap 2011-01-06 21:22:58 -0800 Created attachment 78207 proposed fix I'm not sure if this patch is technically correct. We also had some URL checks prior to r74807, which moved them, and changed their meaning. But all HTTP regression tests pass. Also, I don't really understand what changed in this revision to make similar code so different. It's possible that there are other regressions I don't know about. I only learned about this particular behavior change because Antti mentioned it in bug 43704 comment 23. 330690 2 78207 koivisto 2011-01-07 01:45:44 -0800 Comment on attachment 78207 proposed fix r74807 did not introduce that test (the hash was called m_reloadedURLs before it). This is a long standing webkit behavior. You will need to provide a much better justification for changing this. 330709 3 koivisto 2011-01-07 02:40:41 -0800 Ah, there is an actual regression with no-store (that's what you are testing, the current bug title says no-cache). The old code would never use no-store resources from cache (except when CachePolicy was CachePolicyHistoryBuffer) due to a check in MemoryCache::requestResource(). The correct fix is to add that case to CachedResourceLoader::determineRevalidationPolicy() just before the m_validatedURLs check. 330776 4 ap 2011-01-07 08:34:18 -0800 > This is a long standing webkit behavior. What exactly is the behavior? It's untested, and I can't even tell what to test for. > The correct fix is to add that case to CachedResourceLoader::determineRevalidationPolicy() just before the m_validatedURLs check. Would you be willing to make the fix? 330785 5 ap 2011-01-07 08:50:38 -0800 You are right that I was confused about no-store vs. no-cache here. However, no-cache seems to be in need of fixing, too. Firefox always revalidates these. Perhaps this should be tracked in a separate bug, but my patch takes care of both. 330793 6 ap 2011-01-07 09:03:45 -0800 The Firefox behavior is of course to re-fetch no-store resources with unconditional requests, and to revalidate no-cache ones. Do we want to match that exactly? I think that we do. This is the same observed behavior that's achieved with my patch. It needs more extensive tests, but what do you think about the code change? 330796 7 koivisto 2011-01-07 09:10:46 -0800 (In reply to comment #5) > You are right that I was confused about no-store vs. no-cache here. > > However, no-cache seems to be in need of fixing, too. Firefox always revalidates these. Perhaps this should be tracked in a separate bug, but my patch takes care of both. Yeah, I can patch this for no-store. Surprised we didn't have a test for that. Since we have managed with not revalidating no-cache resources so far I think we should stick with that behavior. If people want their loads to reach server they can use no-store. It would be good to get test cases though. Tests should also differentiate between regular page load and what happens afterwards with timer initiated loads. 330815 8 ap 2011-01-07 09:41:19 -0800 There are two reasons why I would like to fix our no-cache behavior, too: 1) Practical Firefox compatibility (well, I don't have actual examples of broken sites, but matching Firefox is what we usually do for cache behavior). 2) I don't wan't to keep confusing authors about the meaning of no-cache. This is difficult for them, so sticking to the spec and to Firefox will be helpful, and may make a site or two more robust. 3) A reason why I want to eliminate this check altogether is that URL is never a sufficient resource identifier, so all code that's using it alone is wrong. A resource is identified by it URL plus request headers, plus it's considered unique per request if it's no-store, plus the server can always decide to unique it if it's no-cache. My change obviously doesn't get us there, but keeping this check just adds to general confusion. 330816 9 ap 2011-01-07 09:41:46 -0800 > There are two reasons I made that three! :) 330821 10 ap 2011-01-07 09:45:26 -0800 > If people want their loads to reach server they can use no-store. I don't think that's a great option. With no-store, the request would be non-conditional, which is horrible for performance. So, reason #4: 4) We should fix no-cache, because that's pretty common and desirable behavior. Always get fresh data, but don't waste bandwidth if it hasn't changed. 331591 11 78393 koivisto 2011-01-10 05:54:54 -0800 Created attachment 78393 patch 331597 12 koivisto 2011-01-10 06:07:44 -0800 (In reply to comment #10) > I don't think that's a great option. With no-store, the request would be non-conditional, which is horrible for performance. So, reason #4: > > 4) We should fix no-cache, because that's pretty common and desirable behavior. Always get fresh data, but don't waste bandwidth if it hasn't changed. You are probably right (though all expiration should be treated uniformly, no-cache should not be a special case) for what should happen post-load. However we should never load a resource more than once during page loading (with this patch we may in certain cases). 331640 13 78393 ap 2011-01-10 08:48:25 -0800 Comment on attachment 78393 patch View in context: https://bugs.webkit.org/attachment.cgi?id=78393&action=review > However we should never load a resource more than once during page loading Definitely agreed. Would you be willing to file a bug about no-cache? > Source/WebCore/ChangeLog:6 > + REGRESSION(r74807): No-cache is ignored within a document No-cache? > LayoutTests/ChangeLog:10 > + * http/tests/misc/resources/random.php: Added. I'd have renamed this random-no-store.php. Maybe the .html file, too. 331681 14 koivisto 2011-01-10 10:33:51 -0800 http://trac.webkit.org/changeset/75384 331693 15 koivisto 2011-01-10 10:45:16 -0800 (In reply to comment #13) > Would you be willing to file a bug about no-cache? Bug 52153. > > > Source/WebCore/ChangeLog:6 > > + REGRESSION(r74807): No-cache is ignored within a document > > No-cache? > > > LayoutTests/ChangeLog:10 > > + * http/tests/misc/resources/random.php: Added. > > I'd have renamed this random-no-store.php. Maybe the .html file, too. Fixed these. 78207 2011-01-06 21:22:58 -0800 2011-01-10 05:54:54 -0800 proposed fix SameURL.txt text/plain 4662 ap SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvQ2hhbmdlTG9n CShyZXZpc2lvbiA3NTIyNykKKysrIFdlYkNvcmUvQ2hhbmdlTG9nCSh3b3JraW5nIGNvcHkpCkBA IC0xLDMgKzEsMTYgQEAKKzIwMTEtMDEtMDYgIEFsZXhleSBQcm9za3VyeWFrb3YgIDxhcEBhcHBs ZS5jb20+CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAgICAg aHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTUyMDQ0CisgICAgICAgIFJF R1JFU1NJT04ocjc0ODA3KTogTm8tY2FjaGUgaXMgaWdub3JlZCB3aXRoaW4gYSBkb2N1bWVudAor CisgICAgICAgIFRlc3Q6IGh0dHAvdGVzdHMvbWlzYy9tdWx0aXBsZS11bmNhY2hlZC1pbnN0YW5j ZXMuaHRtbAorCisgICAgICAgICogbG9hZGVyL2NhY2hlL0NhY2hlZFJlc291cmNlTG9hZGVyLmNw cDoKKyAgICAgICAgKFdlYkNvcmU6OkNhY2hlZFJlc291cmNlTG9hZGVyOjpkZXRlcm1pbmVSZXZh bGlkYXRpb25Qb2xpY3kpOgorICAgICAgICBJZGVudGljYWwgcmVzb3VyY2UgVVJMIGhhcyBub3Ro aW5nIHRvIGRvIHdpdGggd2hldGhlciB3ZSBzaG91bGQgdXNlIGNhY2hlZCBkYXRhLgorCiAyMDEx LTAxLTA2ICBDaHJpcyBNYXJyaW4gIDxjbWFycmluQGFwcGxlLmNvbT4KIAogICAgICAgICBVbnJl dmlld2VkLgpJbmRleDogV2ViQ29yZS9sb2FkZXIvY2FjaGUvQ2FjaGVkUmVzb3VyY2VMb2FkZXIu Y3BwCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT0KLS0tIFdlYkNvcmUvbG9hZGVyL2NhY2hlL0NhY2hlZFJlc291cmNlTG9h ZGVyLmNwcAkocmV2aXNpb24gNzUyMjQpCisrKyBXZWJDb3JlL2xvYWRlci9jYWNoZS9DYWNoZWRS ZXNvdXJjZUxvYWRlci5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTQwMiwxMCArNDAyLDYgQEAgQ2Fj aGVkUmVzb3VyY2VMb2FkZXI6OlJldmFsaWRhdGlvblBvbGljeQogICAgIGlmIChleGlzdGluZ1Jl c291cmNlLT5pc1ByZWxvYWRlZCgpKQogICAgICAgICByZXR1cm4gVXNlOwogICAgIAotICAgIC8v IEF2b2lkIGxvYWRpbmcgdGhlIHNhbWUgcmVzb3VyY2UgbXVsdGlwbGUgdGltZXMgZm9yIGEgc2lu Z2xlIGRvY3VtZW50LCBldmVuIGlmIHRoZSBjYWNoZSBwb2xpY2llcyB3b3VsZCB0ZWxsIHVzIHRv LgotICAgIGlmIChtX3ZhbGlkYXRlZFVSTHMuY29udGFpbnMoZXhpc3RpbmdSZXNvdXJjZS0+dXJs KCkpKQotICAgICAgICByZXR1cm4gVXNlOwotICAgIAogICAgIC8vIENhY2hlUG9saWN5UmVsb2Fk IGFsd2F5cyByZWxvYWRzCiAgICAgaWYgKGNhY2hlUG9saWN5KCkgPT0gQ2FjaGVQb2xpY3lSZWxv YWQpIHsKICAgICAgICAgTE9HKFJlc291cmNlTG9hZGluZywgIkNhY2hlZFJlc291cmNlTG9hZGVy OjpkZXRlcm1pbmVSZXZhbGlkYXRpb25Qb2xpY3kgcmVsb2FkaW5nIGR1ZSB0byBDYWNoZVBvbGlj eVJlbG9hZC4iKTsKSW5kZXg6IExheW91dFRlc3RzL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBM YXlvdXRUZXN0cy9DaGFuZ2VMb2cJKHJldmlzaW9uIDc1MjI3KQorKysgTGF5b3V0VGVzdHMvQ2hh bmdlTG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEsMTQgQEAKKzIwMTEtMDEtMDYgIEFsZXhl eSBQcm9za3VyeWFrb3YgIDxhcEBhcHBsZS5jb20+CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9C T0RZIChPT1BTISkuCisKKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcu Y2dpP2lkPTUyMDQ0CisgICAgICAgIFJFR1JFU1NJT04ocjc0ODA3KTogTm8tY2FjaGUgaXMgaWdu b3JlZCB3aXRoaW4gYSBkb2N1bWVudAorCisgICAgICAgICogaHR0cC90ZXN0cy9taXNjL211bHRp cGxlLXVuY2FjaGVkLWluc3RhbmNlcy1leHBlY3RlZC50eHQ6IEFkZGVkLgorICAgICAgICAqIGh0 dHAvdGVzdHMvbWlzYy9tdWx0aXBsZS11bmNhY2hlZC1pbnN0YW5jZXMuaHRtbDogQWRkZWQuCisg ICAgICAgICogaHR0cC90ZXN0cy9taXNjL3Jlc291cmNlcy9yYW5kb20ucGhwOiBBZGRlZC4KKwog MjAxMS0wMS0wNiAgU3RlcGhlbiBXaGl0ZSAgPHNlbm9yYmxhbmNvQGNocm9taXVtLm9yZz4KIAog ICAgICAgICBVbnJldmlld2VkOyBjaHJvbWl1bSB0ZXN0IGV4cGVjdGF0aW9ucyB1cGRhdGUuCklu ZGV4OiBMYXlvdXRUZXN0cy9odHRwL3Rlc3RzL21pc2MvbXVsdGlwbGUtdW5jYWNoZWQtaW5zdGFu Y2VzLWV4cGVjdGVkLnR4dAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBMYXlvdXRUZXN0cy9odHRwL3Rlc3RzL21p c2MvbXVsdGlwbGUtdW5jYWNoZWQtaW5zdGFuY2VzLWV4cGVjdGVkLnR4dAkocmV2aXNpb24gMCkK KysrIExheW91dFRlc3RzL2h0dHAvdGVzdHMvbWlzYy9tdWx0aXBsZS11bmNhY2hlZC1pbnN0YW5j ZXMtZXhwZWN0ZWQudHh0CShyZXZpc2lvbiAwKQpAQCAtMCwwICsxLDIgQEAKK1BBU1MKKwoKUHJv cGVydHkgY2hhbmdlcyBvbjogTGF5b3V0VGVzdHMvaHR0cC90ZXN0cy9taXNjL211bHRpcGxlLXVu Y2FjaGVkLWluc3RhbmNlcy1leHBlY3RlZC50eHQKX19fX19fX19fX19fX19fX19fX19fX19fX19f X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fXwpBZGRlZDogc3ZuOm1pbWUt dHlwZQogICArIHRleHQvcGxhaW4KQWRkZWQ6IHN2bjplb2wtc3R5bGUKICAgKyBuYXRpdmUKCklu ZGV4OiBMYXlvdXRUZXN0cy9odHRwL3Rlc3RzL21pc2MvbXVsdGlwbGUtdW5jYWNoZWQtaW5zdGFu Y2VzLmh0bWwKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVzdHMvaHR0cC90ZXN0cy9taXNjL211bHRp cGxlLXVuY2FjaGVkLWluc3RhbmNlcy5odG1sCShyZXZpc2lvbiAwKQorKysgTGF5b3V0VGVzdHMv aHR0cC90ZXN0cy9taXNjL211bHRpcGxlLXVuY2FjaGVkLWluc3RhbmNlcy5odG1sCShyZXZpc2lv biAwKQpAQCAtMCwwICsxLDMwIEBACis8Ym9keT4KKzxzY3JpcHQ+Cit2YXIgY291bnQgPSAwOwor CitpZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKSB7CisgICAgbGF5b3V0VGVzdENvbnRy b2xsZXIuZHVtcEFzVGV4dCgpOworICAgIGxheW91dFRlc3RDb250cm9sbGVyLndhaXRVbnRpbERv bmUoKTsKK30KKworZnVuY3Rpb24gYWRkUmFuZG9tKCkKK3sKKyAgICB2YXIgc2NyaXB0ID0gZG9j dW1lbnQuY3JlYXRlRWxlbWVudCgic2NyaXB0Iik7CisgICAgc2NyaXB0LnNldEF0dHJpYnV0ZSgi c3JjIiwgInJlc291cmNlcy9yYW5kb20ucGhwIik7CisgICAgZG9jdW1lbnQuYm9keS5hcHBlbmRD aGlsZChzY3JpcHQpOworfQorKytjb3VudDsKK2FkZFJhbmRvbSgpOworCitzZXRJbnRlcnZhbChm dW5jdGlvbigpIHsKKyAgICBpZiAoY291bnQgPT0gMSAmJiBkb2N1bWVudC5nZXRFbGVtZW50c0J5 VGFnTmFtZSgicCIpLmxlbmd0aCA9PSAxKSB7CisgICAgICAgICsrY291bnQ7CisgICAgICAgIGFk ZFJhbmRvbSgpOworICAgIH0gZWxzZSBpZiAoZG9jdW1lbnQuZ2V0RWxlbWVudHNCeVRhZ05hbWUo InAiKS5sZW5ndGggPT0gMikgeworICAgICAgICBkb2N1bWVudC53cml0ZShkb2N1bWVudC5nZXRF bGVtZW50c0J5VGFnTmFtZSgicCIpWzBdLmlubmVySFRNTCA9PSBkb2N1bWVudC5nZXRFbGVtZW50 c0J5VGFnTmFtZSgicCIpWzFdLmlubmVySFRNTCA/IAorICAgICAgICAgICAgIkZBSUw8YnI+IiA6 ICJQQVNTPGJyPiIpOworICAgICAgICBpZiAod2luZG93LmxheW91dFRlc3RDb250cm9sbGVyKQor ICAgICAgICAgICAgbGF5b3V0VGVzdENvbnRyb2xsZXIubm90aWZ5RG9uZSgpOworICAgIH0KK30s IDEwKTsKKzwvc2NyaXB0PgoKUHJvcGVydHkgY2hhbmdlcyBvbjogTGF5b3V0VGVzdHMvaHR0cC90 ZXN0cy9taXNjL211bHRpcGxlLXVuY2FjaGVkLWluc3RhbmNlcy5odG1sCl9fX19fX19fX19fX19f X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX18KQWRk ZWQ6IHN2bjptaW1lLXR5cGUKICAgKyB0ZXh0L2h0bWwKCkluZGV4OiBMYXlvdXRUZXN0cy9odHRw L3Rlc3RzL21pc2MvcmVzb3VyY2VzL3JhbmRvbS5waHAKPT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0VGVz dHMvaHR0cC90ZXN0cy9taXNjL3Jlc291cmNlcy9yYW5kb20ucGhwCShyZXZpc2lvbiAwKQorKysg TGF5b3V0VGVzdHMvaHR0cC90ZXN0cy9taXNjL3Jlc291cmNlcy9yYW5kb20ucGhwCShyZXZpc2lv biAwKQpAQCAtMCwwICsxLDkgQEAKKzw/cGhwCisgICAgaGVhZGVyKCJDYWNoZS1Db250cm9sOiBu by1zdG9yZSIpOworPz4KK3ZhciBwID0gZG9jdW1lbnQuY3JlYXRlRWxlbWVudCgicCIpOworcC5h cHBlbmRDaGlsZChkb2N1bWVudC5jcmVhdGVUZXh0Tm9kZSgiPD9waHAKKyAgICBwcmludCByYW5k KCk7Cis/PgorIikpOworZG9jdW1lbnQuYm9keS5hcHBlbmRDaGlsZChwKTsK 78393 2011-01-10 05:54:54 -0800 2011-01-10 08:48:25 -0800 patch no-store.patch text/plain 5178 koivisto SW5kZXg6IFNvdXJjZS9XZWJDb3JlL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2Vi Q29yZS9DaGFuZ2VMb2cJKHJldmlzaW9uIDc1MzY2KQorKysgU291cmNlL1dlYkNvcmUvQ2hhbmdl TG9nCSh3b3JraW5nIGNvcHkpCkBAIC0xLDMgKzEsMTkgQEAKKzIwMTEtMDEtMTAgIEFudHRpIEtv aXZpc3RvICA8YW50dGlAYXBwbGUuY29tPgorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQub3JnL3Nob3dfYnVnLmNnaT9p ZD01MjA0NAorICAgICAgICBSRUdSRVNTSU9OKHI3NDgwNyk6IE5vLWNhY2hlIGlzIGlnbm9yZWQg d2l0aGluIGEgZG9jdW1lbnQKKyAgICAgICAgCisgICAgICAgIERvbid0IHVzZSBjYWNoZWQgcmVz b3VyY2VzIHdpdGggQ2FjaGUtY29udHJvbDogbm8tc3RvcmUgKHVubGVzcyBjYWNoZSBwb2xpY3kg aXMgQ2FjaGVQb2xpY3lIaXN0b3J5QnVmZmVyKS4KKyAgICAgICAgVGhpcyBtYXRjaGVzIGEgYmVo YXZpb3IgdGhhdCBnb3QgbG9zdCBpbiByNzQ4MDcuCisKKyAgICAgICAgVGVzdDogaHR0cC90ZXN0 cy9taXNjL211bHRpcGxlLXVuY2FjaGVkLWluc3RhbmNlcy5odG1sCisgICAgICAgIChieSBBbGV4 ZXkgUHJvc2t1cnlha292KQorCisgICAgICAgICogbG9hZGVyL2NhY2hlL0NhY2hlZFJlc291cmNl TG9hZGVyLmNwcDoKKyAgICAgICAgKFdlYkNvcmU6OkNhY2hlZFJlc291cmNlTG9hZGVyOjpkZXRl cm1pbmVSZXZhbGlkYXRpb25Qb2xpY3kpOgorCiAyMDExLTAxLTEwICBJbHlhIFRpa2hvbm92c2t5 ICA8bG9pc2xvQGNocm9taXVtLm9yZz4KIAogICAgICAgICBSZXZpZXdlZCBieSBQYXZlbCBGZWxk bWFuLgpJbmRleDogU291cmNlL1dlYkNvcmUvbG9hZGVyL2NhY2hlL0NhY2hlZFJlc291cmNlTG9h ZGVyLmNwcAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvV2ViQ29yZS9sb2FkZXIvY2FjaGUvQ2FjaGVk UmVzb3VyY2VMb2FkZXIuY3BwCShyZXZpc2lvbiA3NTM2NikKKysrIFNvdXJjZS9XZWJDb3JlL2xv YWRlci9jYWNoZS9DYWNoZWRSZXNvdXJjZUxvYWRlci5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTQw Miw2ICs0MDIsMTYgQEAgQ2FjaGVkUmVzb3VyY2VMb2FkZXI6OlJldmFsaWRhdGlvblBvbGljeQog ICAgIGlmIChleGlzdGluZ1Jlc291cmNlLT5pc1ByZWxvYWRlZCgpKQogICAgICAgICByZXR1cm4g VXNlOwogICAgIAorICAgIC8vIENhY2hlUG9saWN5SGlzdG9yeUJ1ZmZlciB1c2VzIHRoZSBjYWNo ZSBubyBtYXR0ZXIgd2hhdC4KKyAgICBpZiAoY2FjaGVQb2xpY3koKSA9PSBDYWNoZVBvbGljeUhp c3RvcnlCdWZmZXIpCisgICAgICAgIHJldHVybiBVc2U7CisKKyAgICAvLyBOZXZlciByZXVzZSBy ZXNvdXJjZXMgd2l0aCBDYWNoZS1jb250cm9sOiBuby1zdG9yZS4KKyAgICBpZiAoZXhpc3RpbmdS ZXNvdXJjZS0+cmVzcG9uc2UoKS5jYWNoZUNvbnRyb2xDb250YWluc05vU3RvcmUoKSkgeworICAg ICAgICBMT0coUmVzb3VyY2VMb2FkaW5nLCAiQ2FjaGVkUmVzb3VyY2VMb2FkZXI6OmRldGVybWlu ZVJldmFsaWRhdGlvblBvbGljeSByZWxvYWRpbmcgZHVlIHRvIENhY2hlLWNvbnRyb2w6IG5vLXN0 b3JlLiIpOworICAgICAgICByZXR1cm4gUmVsb2FkOworICAgIH0KKwogICAgIC8vIEF2b2lkIGxv YWRpbmcgdGhlIHNhbWUgcmVzb3VyY2UgbXVsdGlwbGUgdGltZXMgZm9yIGEgc2luZ2xlIGRvY3Vt ZW50LCBldmVuIGlmIHRoZSBjYWNoZSBwb2xpY2llcyB3b3VsZCB0ZWxsIHVzIHRvLgogICAgIGlm IChtX3ZhbGlkYXRlZFVSTHMuY29udGFpbnMoZXhpc3RpbmdSZXNvdXJjZS0+dXJsKCkpKQogICAg ICAgICByZXR1cm4gVXNlOwpAQCAtNDEyLDEwICs0MjIsNiBAQCBDYWNoZWRSZXNvdXJjZUxvYWRl cjo6UmV2YWxpZGF0aW9uUG9saWN5CiAgICAgICAgIHJldHVybiBSZWxvYWQ7CiAgICAgfQogICAg IAotICAgIC8vIENhY2hlUG9saWN5SGlzdG9yeUJ1ZmZlciB1c2VzIHRoZSBjYWNoZSBubyBtYXR0 ZXIgd2hhdC4KLSAgICBpZiAoY2FjaGVQb2xpY3koKSA9PSBDYWNoZVBvbGljeUhpc3RvcnlCdWZm ZXIpCi0gICAgICAgIHJldHVybiBVc2U7Ci0gICAgCiAgICAgLy8gV2UnbGwgdHJ5IHRvIHJlbG9h ZCB0aGUgcmVzb3VyY2UgaWYgaXQgZmFpbGVkIGxhc3QgdGltZS4KICAgICBpZiAoZXhpc3RpbmdS ZXNvdXJjZS0+ZXJyb3JPY2N1cnJlZCgpKSB7CiAgICAgICAgIExPRyhSZXNvdXJjZUxvYWRpbmcs ICJDYWNoZWRSZXNvdXJjZUxvYWRlcjo6ZGV0ZXJtaW5lUmV2YWxpZGF0aW9uUG9saWN5ZSByZWxv YWRpbmcgZHVlIHRvIHJlc291cmNlIGJlaW5nIGluIHRoZSBlcnJvciBzdGF0ZSIpOwpJbmRleDog TGF5b3V0VGVzdHMvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIExheW91dFRlc3RzL0NoYW5nZUxv ZwkocmV2aXNpb24gNzUzNjYpCisrKyBMYXlvdXRUZXN0cy9DaGFuZ2VMb2cJKHdvcmtpbmcgY29w eSkKQEAgLTEsMyArMSwxNCBAQAorMjAxMS0wMS0xMCAgQW50dGkgS29pdmlzdG8gIDxhbnR0aUBh cHBsZS5jb20+CisKKyAgICAgICAgUmV2aWV3ZWQgYnkgTk9CT0RZIChPT1BTISkuCisKKyAgICAg ICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTUyMDQ0CisgICAgICAg IFJFR1JFU1NJT04ocjc0ODA3KTogTm8tY2FjaGUgaXMgaWdub3JlZCB3aXRoaW4gYSBkb2N1bWVu dAorCisgICAgICAgICogaHR0cC90ZXN0cy9taXNjL211bHRpcGxlLXVuY2FjaGVkLWluc3RhbmNl cy1leHBlY3RlZC50eHQ6IEFkZGVkLgorICAgICAgICAqIGh0dHAvdGVzdHMvbWlzYy9tdWx0aXBs ZS11bmNhY2hlZC1pbnN0YW5jZXMuaHRtbDogQWRkZWQuCisgICAgICAgICogaHR0cC90ZXN0cy9t aXNjL3Jlc291cmNlcy9yYW5kb20ucGhwOiBBZGRlZC4KKwogMjAxMS0wMS0wOSAgSnVzdGluIEdh cmNpYSAgPGp1c3Rpbi5nYXJjaWFAYXBwbGUuY29tPgogCiAgICAgICAgIFVucmV2aWV3ZWQuICBB ZGRpbmcgc3ZnL3RleHQvY2FyZXQtaW4tc3ZnLXRleHQueGh0bWwgdG8gU2tpcHBlZCBsaXN0IGZv ciBndGssIHdoZXJlIHRoZXJlIApJbmRleDogTGF5b3V0VGVzdHMvaHR0cC90ZXN0cy9taXNjL211 bHRpcGxlLXVuY2FjaGVkLWluc3RhbmNlcy1leHBlY3RlZC50eHQKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5 b3V0VGVzdHMvaHR0cC90ZXN0cy9taXNjL211bHRpcGxlLXVuY2FjaGVkLWluc3RhbmNlcy1leHBl Y3RlZC50eHQJKHJldmlzaW9uIDApCisrKyBMYXlvdXRUZXN0cy9odHRwL3Rlc3RzL21pc2MvbXVs dGlwbGUtdW5jYWNoZWQtaW5zdGFuY2VzLWV4cGVjdGVkLnR4dAkocmV2aXNpb24gMCkKQEAgLTAs MCArMSwyIEBACitQQVNTCisKSW5kZXg6IExheW91dFRlc3RzL2h0dHAvdGVzdHMvbWlzYy9tdWx0 aXBsZS11bmNhY2hlZC1pbnN0YW5jZXMuaHRtbAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBMYXlvdXRUZXN0cy9o dHRwL3Rlc3RzL21pc2MvbXVsdGlwbGUtdW5jYWNoZWQtaW5zdGFuY2VzLmh0bWwJKHJldmlzaW9u IDApCisrKyBMYXlvdXRUZXN0cy9odHRwL3Rlc3RzL21pc2MvbXVsdGlwbGUtdW5jYWNoZWQtaW5z dGFuY2VzLmh0bWwJKHJldmlzaW9uIDApCkBAIC0wLDAgKzEsMzAgQEAKKzxib2R5PgorPHNjcmlw dD4KK3ZhciBjb3VudCA9IDA7CisKK2lmICh3aW5kb3cubGF5b3V0VGVzdENvbnRyb2xsZXIpIHsK KyAgICBsYXlvdXRUZXN0Q29udHJvbGxlci5kdW1wQXNUZXh0KCk7CisgICAgbGF5b3V0VGVzdENv bnRyb2xsZXIud2FpdFVudGlsRG9uZSgpOworfQorCitmdW5jdGlvbiBhZGRSYW5kb20oKQorewor ICAgIHZhciBzY3JpcHQgPSBkb2N1bWVudC5jcmVhdGVFbGVtZW50KCJzY3JpcHQiKTsKKyAgICBz Y3JpcHQuc2V0QXR0cmlidXRlKCJzcmMiLCAicmVzb3VyY2VzL3JhbmRvbS5waHAiKTsKKyAgICBk b2N1bWVudC5ib2R5LmFwcGVuZENoaWxkKHNjcmlwdCk7Cit9CisrK2NvdW50OworYWRkUmFuZG9t KCk7CisKK3NldEludGVydmFsKGZ1bmN0aW9uKCkgeworICAgIGlmIChjb3VudCA9PSAxICYmIGRv Y3VtZW50LmdldEVsZW1lbnRzQnlUYWdOYW1lKCJwIikubGVuZ3RoID09IDEpIHsKKyAgICAgICAg Kytjb3VudDsKKyAgICAgICAgYWRkUmFuZG9tKCk7CisgICAgfSBlbHNlIGlmIChkb2N1bWVudC5n ZXRFbGVtZW50c0J5VGFnTmFtZSgicCIpLmxlbmd0aCA9PSAyKSB7CisgICAgICAgIGRvY3VtZW50 LndyaXRlKGRvY3VtZW50LmdldEVsZW1lbnRzQnlUYWdOYW1lKCJwIilbMF0uaW5uZXJIVE1MID09 IGRvY3VtZW50LmdldEVsZW1lbnRzQnlUYWdOYW1lKCJwIilbMV0uaW5uZXJIVE1MID8gCisgICAg ICAgICAgICAiRkFJTDxicj4iIDogIlBBU1M8YnI+Iik7CisgICAgICAgIGlmICh3aW5kb3cubGF5 b3V0VGVzdENvbnRyb2xsZXIpCisgICAgICAgICAgICBsYXlvdXRUZXN0Q29udHJvbGxlci5ub3Rp ZnlEb25lKCk7CisgICAgfQorfSwgMTApOworPC9zY3JpcHQ+CkluZGV4OiBMYXlvdXRUZXN0cy9o dHRwL3Rlc3RzL21pc2MvcmVzb3VyY2VzL3JhbmRvbS5waHAKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gTGF5b3V0 VGVzdHMvaHR0cC90ZXN0cy9taXNjL3Jlc291cmNlcy9yYW5kb20ucGhwCShyZXZpc2lvbiAwKQor KysgTGF5b3V0VGVzdHMvaHR0cC90ZXN0cy9taXNjL3Jlc291cmNlcy9yYW5kb20ucGhwCShyZXZp c2lvbiAwKQpAQCAtMCwwICsxLDkgQEAKKzw/cGhwCisgICAgaGVhZGVyKCJDYWNoZS1Db250cm9s OiBuby1zdG9yZSIpOworPz4KK3ZhciBwID0gZG9jdW1lbnQuY3JlYXRlRWxlbWVudCgicCIpOwor cC5hcHBlbmRDaGlsZChkb2N1bWVudC5jcmVhdGVUZXh0Tm9kZSgiPD9waHAKKyAgICBwcmludCBy YW5kKCk7Cis/PgorIikpOworZG9jdW1lbnQuYm9keS5hcHBlbmRDaGlsZChwKTsK