We are dying in the following function: t = function() { var e = g.defer(), a = this, f = a.config.appId; n.info.always("Initializing Cast SDK - Version: " + NFGCKFrameworkVersion + " - AppId: " + f); var h = c.processIflist(nrdp.device.iflist); a.networkId = c.networkIdFromInterface(h), l.forEach(o, function(b) { b.handler = b.handler.bind(a), nrdp.device.addEventListener(b.name, b.handler) }); var i = NFGCKFilterCriteria.criteriaForAvailableApplicationWithID(f); return NFGCKDeviceScanner.initWithFilterCriteria(i, function(c) { return c ? (d = c, b = NFGCKDeviceScannerListener.init(), b.didComeOnline = v.bind(a), b.didGoOffline = r.bind(a), b.didChange = s.bind(a), d.addListener(b), d.startScan(), void e.resolve()) : void e.reject(new m(m.ERRORS.START_ERROR, "No cast scanner instance")) }), e.promise } We are dying because there is no initWithFilterCriteria property on NFGCKDeviceScanner. This looks like this is an inadvertent result of change set <http://trac.webkit.org/changeset/200114>.
<rdar://problem/26878852>
Created attachment 282401 [details] Patch
Comment on attachment 282401 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=282401&action=review I would really like an explanation of the actual bug in the ChangeLog or in Radar. I have an hypothesis as to why the dyld_get_program_sdk_version() was failing but I can't find if my hypothesis holds from this changelog or radar. This should be documented IMHO. > Source/JavaScriptCore/API/JSWrapperMap.mm:665 > + static int32_t versionOfLinkTimeJavaScriptCore = 0; > + if (!versionOfLinkTimeJavaScriptCore) Looks like something that should use dispatch_once(). > Source/JavaScriptCore/API/JSWrapperMap.mm:667 > + // Only do the link time version comparision if we linked directly with JavaScriptCore Typo: comparision
(In reply to comment #3) > Comment on attachment 282401 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=282401&action=review > > I would really like an explanation of the actual bug in the ChangeLog or in > Radar. Added. > I have an hypothesis as to why the dyld_get_program_sdk_version() was > failing but I can't find if my hypothesis holds from this changelog or > radar. This should be documented IMHO. > > > Source/JavaScriptCore/API/JSWrapperMap.mm:665 > > + static int32_t versionOfLinkTimeJavaScriptCore = 0; > > + if (!versionOfLinkTimeJavaScriptCore) > > Looks like something that should use dispatch_once(). > > > Source/JavaScriptCore/API/JSWrapperMap.mm:667 > > + // Only do the link time version comparision if we linked directly with JavaScriptCore > > Typo: comparision Fixed.
Committed r202670: <http://trac.webkit.org/changeset/202670>