138690 – Support multiple signatures of diagnostic logging.

RESOLVED FIXED 138690

Support multiple signatures of diagnostic logging.

https://bugs.webkit.org/show_bug.cgi?id=138690

Summary Support multiple signatures of diagnostic logging.

Jer Noble

Reported 2014-11-13 01:02:04 PST

Support multiple signatures of diagnostic logging.

Attachments
Patch (39.25 KB, patch) 2014-11-13 01:11 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (71.77 KB, patch) 2014-11-14 09:01 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (72.54 KB, patch) 2014-11-14 13:26 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (44.25 KB, patch) 2014-11-14 14:18 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (73.87 KB, patch) 2014-11-14 15:06 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (74.55 KB, patch) 2014-11-14 15:38 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (76.93 KB, patch) 2014-11-17 11:24 PST, Jer Noble	no flags	Details Formatted Diff Diff
Patch (76.97 KB, patch) 2014-11-21 15:58 PST, Jer Noble	andersca: review+	Details Formatted Diff Diff
Patch (77.31 KB, patch) 2014-11-21 16:43 PST, Jer Noble	no flags	Details Formatted Diff Diff
Show Obsolete (7) View All Add attachment proposed patch, testcase, etc.

Jer Noble

Comment 1 2014-11-13 01:11:31 PST

Created attachment 241477 [details] Patch

Anders Carlsson

Comment 2 2014-11-13 08:26:44 PST

I think this client should live on MainFrame instead of Page. Darin, what do you think?

Darin Adler

Comment 3 2014-11-13 20:30:36 PST

I agree. Almost anything new we think about adding to Page we could instead add to MainFrame.

Jer Noble

Comment 4 2014-11-13 21:31:32 PST

Frame/MainFrame doesn't seem to have the same kind of "client" infrastructure as page does. In fact, it looks like Page passes its own loaderClientForMainFrame into MainFrame when it creates it. Should MainFrame take a MainFrameClients struct (I guess passed into it inside of or part of PageClients) in its constructor? And what makes this all weirder is that WebPage is created first, then Page, then MainFrame, then WebFrame. So if a client is going to exist at construction time, it's got to be WebPage which creates it. We could have WebFrame set its client on MainFrame after they're all constructed, but that might result in logging messages getting dropped if we add logs which fire in between when Page is constructed and when WebFrame sets its clients on it. And WebFrame currently doesn't have any of its own clients. It's given a client by WebPage. So I guess I don't understand what makes MainFrame the best place to put these new clients?

Jer Noble

Comment 5 2014-11-13 21:42:51 PST

// FIXME: Rename this to PageConfiguration and move it to its own class. struct PageClients { ... }; This would be a lot easier if that FIXME was fixed. As it is, the class-member-class PageClients can't be forward declared.

Jer Noble

Comment 6 2014-11-14 09:01:44 PST

Created attachment 241592 [details] Patch

Jer Noble

Comment 7 2014-11-14 13:26:57 PST

Created attachment 241619 [details] Patch

WebKit Commit Bot

Comment 8 2014-11-14 13:29:05 PST

Attachment 241619 [details] did not pass style-queue: ERROR: Source/WebCore/WebCore.exp.in:0: Source/WebCore/WebCore.exp.in should be sorted, use Tools/Scripts/sort-export-file script [list/order] [5] Total errors found: 1 in 27 files If any of these errors are false positives, please file a bug against check-webkit-style.

Jer Noble

Comment 9 2014-11-14 14:18:52 PST

Created attachment 241624 [details] Patch

Jer Noble

Comment 10 2014-11-14 15:06:29 PST

Created attachment 241629 [details] Patch

Jer Noble

Comment 11 2014-11-14 15:38:31 PST

Created attachment 241634 [details] Patch

Jer Noble

Comment 12 2014-11-17 11:24:18 PST

Created attachment 241728 [details] Patch