Fix debug build GTK assert in TabContentsDragSource::OnDragBegin().

Fix debug build GTK assert in TabContentsDragSource::OnDragBegin().

The assert is:
Gtk: gtk_widget_set_colormap: assertion `!gtk_widget_get_realized
(widget)' failed.

BUG=none
TEST=Navigate to google.com and drag the logo twice.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=54851

[dcheng, 2010-08-03 21:18:33]

I'm not sure if this is the proper fix, since I'm not familiar with GTK.

[Evan Martin, 2010-08-03 21:22:30]

On 2010/08/03 21:18:33, dcheng wrote:
> I'm not sure if this is the proper fix, since I'm not familiar with GTK.

What is the call stack of the assertion?

[dcheng, 2010-08-03 21:28:07]

[28661:28661:435700960538:FATAL:chrome/app/chrome_dll_main.cc(251)] Gtk:
gtk_widget_set_colormap: assertion `!gtk_widget_get_realized (widget)'
failed
Backtrace:
StackTrace::StackTrace() [0x1a5d6a6]
logging::LogMessage::~LogMessage() [0x1a74979]
(anonymous namespace)::GLibLogHandler() [0xf659fe]
0x7fe748536fb9
0x7fe7485373d3
TabContentsDragSource::OnDragBegin() [0x15b5294]
TabContentsDragSource::OnDragBeginThunk() [0x15b5616]
0x7fe748be25de
0x7fe748bf6598
0x7fe748bf7a76
0x7fe748bf7dc8
0x7fe74a7f0e64
TabContentsDragSource::StartDragging() [0x15b488a]
TabContentsViewGtk::StartDragging() [0x137aa3b]
RenderViewHost::OnMsgStartDragging() [0x12ac4e5]
DispatchToMethod<>() [0x12ba65a]
IPC::MessageWithTuple<>::Dispatch<>() [0x12b7909]
RenderViewHost::OnMessageReceived() [0x12a9943]
BrowserRenderProcessHost::OnMessageReceived() [0x163b858]
IPC::ChannelProxy::Context::OnDispatchMessage() [0x2397144]
DispatchToMethod<>() [0x2399c6e]
RunnableMethod<>::Run() [0x23999ac]
MessageLoop::RunTask() [0x1a77355]
MessageLoop::DeferOrRunPendingTask() [0x1a77438]
MessageLoop::DoWork() [0x1a778e4]
base::MessagePumpForUI::HandleDispatch() [0x1ada185]
(anonymous namespace)::WorkSourceDispatch() [0x1ad9667]
0x7fe74852c8c2
0x7fe748530748
0x7fe7485308fc
base::MessagePumpForUI::RunWithDispatcher() [0x1ad9e3a]
base::MessagePumpForUI::Run() [0x1ada5d0]
MessageLoop::RunInternal() [0x1a76b20]
MessageLoop::RunHandler() [0x1a769cc]
MessageLoopForUI::Run() [0x1a77f58]
(anonymous namespace)::RunUIMessageLoop() [0xfdfab5]
BrowserMain() [0xfe19db]
ChromeMain [0xf675fd]
main [0xf684f0]
0x7fe745118c4d
0xf656a9

This happens on the second drag on a page when a drag image has been set.

On Tue, Aug 3, 2010 at 14:22, <evan@chromium.org> wrote:

> On 2010/08/03 21:18:33, dcheng wrote:
>
>> I'm not sure if this is the proper fix, since I'm not familiar with GTK.
>>
>
> What is the call stack of the assertion?
>
>
> http://codereview.chromium.org/3086011/show
>

[Evan Stade, 2010-08-03 21:39:48]

http://codereview.chromium.org/3086011/diff/2001/3001
File chrome/browser/gtk/tab_contents_drag_source.cc (right):

http://codereview.chromium.org/3086011/diff/2001/3001#newcode322
chrome/browser/gtk/tab_contents_drag_source.cc:322: GdkScreen* screen =
gtk_widget_get_screen(drag_icon_);
I think I'd just wrap this block (L322-L325) in a !gtk_widget_get_realized
condition with a comment explaining we only need to do this once.

[dcheng, 2010-08-03 21:59:04]

http://codereview.chromium.org/3086011/diff/2001/3001
File chrome/browser/gtk/tab_contents_drag_source.cc (right):

http://codereview.chromium.org/3086011/diff/2001/3001#newcode322
chrome/browser/gtk/tab_contents_drag_source.cc:322: GdkScreen* screen =
gtk_widget_get_screen(drag_icon_);
On 2010/08/03 21:39:48, Evan Stade wrote:
> I think I'd just wrap this block (L322-L325) in a !gtk_widget_get_realized
> condition with a comment explaining we only need to do this once.

Done.

[Evan Stade, 2010-08-03 22:00:22]

lgtm