This error prints once when opening every page: ERROR: Unhandled web process message 'WebPage:SetPageSuppressed' ../../Source/WebKit2/WebProcess/WebProcess.cpp(654) : virtual void WebKit::WebProcess::didReceiveMessage(IPC::Connection&, IPC::Decoder&) Not sure why the message isn't being handled by WebPage. Maybe it's being received before the WebPage object is created.
(In reply to comment #0) > This error prints once when opening every page: (when opening a new browser tab)
I don't think this is specific to GTK+, the updateThrottleState() can be called too early. Actually, I think the problem is that is called right after the web process is launched, but before the WebPage has registered its message receiver.
Yeah, this is a regression from r207424 I believe.
Gavin, we can't send messages (like updateThrottleState does) before WebPageProxy::initializeWebPage(). Do you think it's safe to move it?
(In reply to comment #4) > Do you think it's safe to move it? Well we should either move it or roll it out. Here's an attempt at the former. It avoids the warning and it *looks* like it's right, but I'm not sure.
Created attachment 295270 [details] Patch
Comment on attachment 295270 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=295270&action=review > Source/WebKit2/UIProcess/WebPageProxy.cpp:-468 > - updateThrottleState(); What’s the reasoning behind removing this one?
Gavin should look!
(In reply to comment #7) > > Source/WebKit2/UIProcess/WebPageProxy.cpp:-468 > > - updateThrottleState(); > > What’s the reasoning behind removing this one? That's in the constructor before the first call to WebPageProxy::initializeWebPage, which only ever occurs in reattachToWebProcess. I moved the calls in that function to be right after initializeWebPage to ensure it's called ASAP. Is it right? I don't know; we don't have an implementation of throttling to test. :(
Seems likely we’d want to do the same thing with updateActivityState. Ultimately I do want Gavin to look at this. Should we land some kind of change in the mean time?
If Gavin's not available to look at it soon, then we should roll out r207699 and r207424.
(In reply to comment #11) > If Gavin's not available to look at it soon, then we should roll out r207699 > and r207424. I looked into this. There are a few more quite significant commits that would need to be rolled out, including bringing back the PageThrottler class. I'll roll them out if Gavin doesn't respond soon (he's been CCed on this for a month now) but it'd certainly be better to avoid doing that.
We should just remove SetPageSuppressed – we've got to the point we can do so. *** This bug has been marked as a duplicate of bug 165158 ***
Super, thanks for handling this properly Gavin!