Google Summer of Code 2013 - GNOME - Archive Integration workspace https://www.lant.com.tw/~lantw44/cgit/cgit.cgi/gsoc2013-epiphany/atom?h=3.7.92 2013-03-20T18:37:56+00:00 2013-03-20T18:37:56+00:00 Xan Lopez xan@igalia.com 2013-03-19T16:04:18+00:00 urn:sha1:cc9e3710001886f6e78c655f237784f4b7bf4ce1 2013-03-20T18:37:56+00:00 Xan Lopez xan@igalia.com 2013-03-20T18:28:53+00:00 urn:sha1:8a8c666da4e43027eed87d93e5c4154fb7b0ff8b It should work now. https://bugzilla.gnome.org/show_bug.cgi?id=695620 2013-03-20T18:37:56+00:00 Xan Lopez xan@igalia.com 2013-03-20T18:28:02+00:00 urn:sha1:f32c375e9b113d5b9668af6ec5379e4e743d01fe Need to do this since we override XDG_DATA_DIRS. https://bugzilla.gnome.org/show_bug.cgi?id=695620 2013-03-20T14:50:07+00:00 Carlos Garcia Campos cgarcia@igalia.com 2013-03-19T18:02:12+00:00 urn:sha1:f65b0e0ef8322139d230c0b5ba60e569ee1f390c As well as the code the prepare the environment for the wen process and web context initialization. This allows to only run that initializations only for the primary instance. It fixes a crash in web process when the ui process finishes too early and makes secondary instances to run faster, since they don't spawn a web process anymore. https://bugzilla.gnome.org/show_bug.cgi?id=696020 2013-03-20T14:04:51+00:00 Xan Lopez xan@igalia.com 2013-03-20T14:03:24+00:00 urn:sha1:8ffda1baa20998c9a1b8be29a3900f5941fb2601 This code is causing some problems because the check happens in a sync call right before a WebView is destroyed. There seems to be a race condition and some times the method call will hang for seconds waiting for a dead process on the other side. We'll figure it out for 3.8.1, but for now disable it since it's causing more harm than good. 2013-03-19T21:47:49+00:00 Gianvito Cavasoli gianvito@gmx.it 2013-03-19T21:47:49+00:00 urn:sha1:694f942afc5c1a2e237921372259b835c98b3122 2013-03-19T19:48:20+00:00 Victor Ibragimov victor.ibragimov@gmail.com 2013-03-19T19:48:08+00:00 urn:sha1:4cbafc06c8ba6cc5533a8a9e0d23181c5b9165a0 2013-03-19T18:45:04+00:00 Bastien Nocera hadess@hadess.net 2013-03-19T18:40:55+00:00 urn:sha1:fd1ce00c45ee40ff22e715f3f45ec49d317c71b1 This guards against error paths that might not return errors. It'll crash in the error path instead :) 2013-03-19T18:45:04+00:00 Bastien Nocera hadess@hadess.net 2013-03-14T09:11:27+00:00 urn:sha1:7eaba5831d5e339ee0ed114ff7478d50f1c255b4 Use atomic operations to manipulate the counter that tracks the number of operations we need to achieve, to ensure coherency when the results are received from different threads. Also use our own main loop rather than spinning on g_main_context_iteration() to avoid CPU burning when waiting for completion. Finally, fix the migrator never returning as the number of operations was not decreased on the ephy_form_auth_data_store() error path. https://bugzilla.gnome.org/show_bug.cgi?id=695828 2013-03-19T12:47:59+00:00 Gustavo Noronha Silva gns@gnome.org 2013-03-16T02:58:09+00:00 urn:sha1:10f300b27a14c450c9a20b12c23f4e542942655c In WebKit2 the back/forward list is not copiable, so that assumption doesn't hold, and the current code path fails. With this change we tell EphyShell to not copy the list and load the URL held by the back item of the existing WebView in the new WebView instead, for WebKit2. https://bugzilla.gnome.org/show_bug.cgi?id=695951