RESOLVED FIXED 26039
[GTK] WebKitWebHistoryItem lifetime fixes 
https://bugs.webkit.org/show_bug.cgi?id=26039
Summary [GTK] WebKitWebHistoryItem lifetime fixes
Xan Lopez
Reported 2009-05-27 00:50:10 PDT
As discussed on IRC the lifetime management of the history items has some problems right now. The attached patches give the details.
Attachments
history1.patch (2.43 KB, patch)
2009-05-27 00:51 PDT, Xan Lopez 		jmalonzo: review+
DetailsFormatted DiffDiff
history2.patch (16.31 KB, patch)
2009-05-27 00:52 PDT, Xan Lopez 		no flags
DetailsFormatted DiffDiff
history2.patch (16.31 KB, patch)
2009-05-27 00:53 PDT, Xan Lopez 		no flags
DetailsFormatted DiffDiff
history3.patch (11.96 KB, patch)
2009-05-27 16:16 PDT, Xan Lopez 		gustavo: review+
DetailsFormatted DiffDiff
Show Obsolete (2) View All Add attachment proposed patch, testcase, etc.
Xan Lopez
Comment 1 2009-05-27 00:51:43 PDT
Created attachment 30698 [details] history1.patch Delete backforward list before page on the webview.
Xan Lopez
Comment 2 2009-05-27 00:52:19 PDT
Created attachment 30699 [details] history2.patch Modify management of history items.
Xan Lopez
Comment 3 2009-05-27 00:53:19 PDT
Created attachment 30700 [details] history2.patch Modify management of history items.
Jan Alonzo
Comment 4 2009-05-27 03:37:43 PDT
Comment on attachment 30698 [details] history1.patch r=me.
Jan Alonzo
Comment 5 2009-05-27 03:56:58 PDT
(In reply to comment #2) > Created an attachment (id=30699) [review] > history2.patch > > Modify management of history items. > The original intention, and why the table is maintained by WebKitWebHistoryItem, was to share the history item between a WebBackForwardList and a *future* WebHistory. It's also the reason why I hesitated to make WebBackForwardList the sole owner of history items (i.e. the hash table maintained by back forward list). But since I don't see WebHistory coming anytime soon, perhaps we can make the WebBackForwardList the owner of history items for now and figure this out once we have a WebHistory? Will that fix this issue?
Xan Lopez
Comment 6 2009-05-27 16:16:12 PDT
Created attachment 30719 [details] history3.patch A slightly less intrusive approach.
Gustavo Noronha (kov)
Comment 7 2009-05-27 16:28:48 PDT
Comment on attachment 30719 [details] history3.patch > + GObjectClass *object_class = G_OBJECT_CLASS(klass); The only problem I could fine is this misplaced * =P
Xan Lopez
Comment 8 2009-05-28 07:54:02 PDT
Landed both patches as 44227 and 44228. Closing.
Note You need to log in before you can comment on or make changes to this bug.