Fix to lose a focus when move the focus from the delayed focus-in/out.

Fix to lose a focus when move the focus from the delayed focus-in/out.

Some applications likes google-chrome URL entry could have the delayed
focus-in and focus-out events and ibus fails to get the new focus.

BUG=rhbz#919863, rhbz#1136623

[Peng, 2013-03-12 18:12:21]

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode694
bus/inputcontext.c:694: */
Sorry. I can not understand below code.

[fujiwara, 2013-03-14 07:00:53]

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode694
bus/inputcontext.c:694: */
On 2013/03/12 18:12:21, Peng wrote:
> Sorry. I can not understand below code.

If I press Alt+Tab and move the focus from gedit to thunderbird, the following
focus events are happened:

1) (gedit:445): IBUS-WARNING **: 15:32:36:717  focus_out
2) (thunderbird:495): IBUS-WARNING **: 15:32:36:866  focus_in
3) (gedit:445): IBUS-WARNING **: 15:32:36:875  focus_in
4) (gedit:445): IBUS-WARNING **: 15:32:36:890  focus_out

1) It seems XI2 Alt+Tab causes focus-out event at first.
2) focused_context is changed from gedit to thunderbird.
3) It seems Alt+Tab causes focus-in event when Alt+Tab UI is closed and
focused_context is changed from thunderbird to gedit.
4) It seems Alt+Tab causes focus-out event finally and focused_context is
changed from gedit to faked_context in ibus global mode.

Then when _ic_process_key_event() is called, the context in thunderbird does not
have the focus and the proxy engine because the faked_context has those.

This patch applies the proxy engine from faked_context to the context if the
context->engine is NULL.

[Peng, 2013-03-14 21:42:44]

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode694
bus/inputcontext.c:694: */
On 2013/03/14 07:00:53, fujiwara wrote:
> On 2013/03/12 18:12:21, Peng wrote:
> > Sorry. I can not understand below code.
> 
> If I press Alt+Tab and move the focus from gedit to thunderbird, the following
> focus events are happened:
> 
> 1) (gedit:445): IBUS-WARNING **: 15:32:36:717  focus_out
> 2) (thunderbird:495): IBUS-WARNING **: 15:32:36:866  focus_in
> 3) (gedit:445): IBUS-WARNING **: 15:32:36:875  focus_in
> 4) (gedit:445): IBUS-WARNING **: 15:32:36:890  focus_out
> 
> 1) It seems XI2 Alt+Tab causes focus-out event at first.
> 2) focused_context is changed from gedit to thunderbird.
> 3) It seems Alt+Tab causes focus-in event when Alt+Tab UI is closed and
> focused_context is changed from thunderbird to gedit.
> 4) It seems Alt+Tab causes focus-out event finally and focused_context is
> changed from gedit to faked_context in ibus global mode.
> 
> Then when _ic_process_key_event() is called, the context in thunderbird does
not
> have the focus and the proxy engine because the faked_context has those.
> 
> This patch applies the proxy engine from faked_context to the context if the
> context->engine is NULL.

Seems 3) should not happen. Could you please check do more research? It is
better to find out why it happens. I am wondering it is caused by ibus im module
or key snooper.

[fujiwara, 2013-03-15 07:54:07]

On 2013/03/14 21:42:44, Peng wrote:
> https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c
> File bus/inputcontext.c (right):
> 
> https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode694
> bus/inputcontext.c:694: */
> On 2013/03/14 07:00:53, fujiwara wrote:
> > On 2013/03/12 18:12:21, Peng wrote:
> > > Sorry. I can not understand below code.
> > 
> > If I press Alt+Tab and move the focus from gedit to thunderbird, the
following
> > focus events are happened:
> > 
> > 1) (gedit:445): IBUS-WARNING **: 15:32:36:717  focus_out
> > 2) (thunderbird:495): IBUS-WARNING **: 15:32:36:866  focus_in
> > 3) (gedit:445): IBUS-WARNING **: 15:32:36:875  focus_in
> > 4) (gedit:445): IBUS-WARNING **: 15:32:36:890  focus_out
> > 
> > 1) It seems XI2 Alt+Tab causes focus-out event at first.
> > 2) focused_context is changed from gedit to thunderbird.
> > 3) It seems Alt+Tab causes focus-in event when Alt+Tab UI is closed and
> > focused_context is changed from thunderbird to gedit.
> > 4) It seems Alt+Tab causes focus-out event finally and focused_context is
> > changed from gedit to faked_context in ibus global mode.
> > 
> > Then when _ic_process_key_event() is called, the context in thunderbird does
> not
> > have the focus and the proxy engine because the faked_context has those.
> > 
> > This patch applies the proxy engine from faked_context to the context if the
> > context->engine is NULL.
> 
> Seems 3) should not happen. Could you please check do more research? It is
> better to find out why it happens. I am wondering it is caused by ibus im
module
> or key snooper.

Probably I think the global keybinding could causes focus in/out and ibus
Ctrl+space also could causes focus in/out.
https://bugzilla.redhat.com/show_bug.cgi?id=909069

This is not caused by ibus im module:

#0  ibus_input_context_focus_in (context=0x22ad3c0) at ibusinputcontext.c:1177
#1  0x00007f7e2addd406 in ibus_im_context_focus_in (context=0x260a170)
    at ibusimcontext.c:825
#2  0x00007f7e3a03bc8d in gtk_im_context_focus_in (context=0x260a170)
    at gtkimcontext.c:539
#3  0x00007f7e3a0400dc in gtk_im_multicontext_focus_in (context=0x25ca990)
    at gtkimmulticontext.c:410
#4  0x00007f7e3a03bc8d in gtk_im_context_focus_in (context=0x25ca990)
    at gtkimcontext.c:539
#5  0x00007f7e3a176c2d in gtk_text_view_focus_in_event (widget=0x25e8020, 
    event=0x20c62f0) at gtktextview.c:5018
#6  0x00007f7e3a0620e4 in _gtk_marshal_BOOLEAN__BOXED (closure=0x2026340, 
    return_value=0x7fff10e8ba30, n_param_values=2, 
    param_values=0x7fff10e8bae0, invocation_hint=0x7fff10e8ba80, 
    marshal_data=0x7f7e3a176b1e <gtk_text_view_focus_in_event>)
    at gtkmarshalers.c:85
#7  0x0000003a78e0fa72 in g_closure_invoke () from /lib64/libgobject-2.0.so.0
#8  0x0000003a78e2119f in signal_emit_unlocked_R ()
   from /lib64/libgobject-2.0.so.0
#9  0x0000003a78e28c34 in g_signal_emit_valist ()
   from /lib64/libgobject-2.0.so.0
#10 0x0000003a78e29702 in g_signal_emit () from /lib64/libgobject-2.0.so.0
#11 0x00007f7e3a20f2f5 in gtk_widget_event_internal (widget=0x25e8020, 
    event=0x20c62f0) at gtkwidget.c:6703
#12 0x00007f7e3a20e944 in gtk_widget_event (widget=0x25e8020, event=0x20c62f0)
    at gtkwidget.c:6360
#13 0x00007f7e3a21f1d3 in gtk_widget_send_focus_change (widget=0x25e8020, 
    event=0x20c62f0) at gtkwidget.c:14451
#14 0x00007f7e3a22d8dd in do_focus_change (widget=0x25e8020, in=1)
    at gtkwindow.c:6154
#15 0x00007f7e3a232fed in window_update_has_focus (window=0x22400c0)
    at gtkwindow.c:9420
#16 0x00007f7e3a233127 in _gtk_window_set_is_active (window=0x22400c0, 
    is_active=1) at gtkwindow.c:9456
#17 0x00007f7e3a22da77 in gtk_window_focus_in_event (widget=0x22400c0, 
    event=0x20c6530) at gtkwindow.c:6208
#18 0x00007f7e3a0620e4 in _gtk_marshal_BOOLEAN__BOXED (closure=0x2026340, 
    return_value=0x7fff10e8c060, n_param_values=2, 
    param_values=0x7fff10e8c110, invocation_hint=0x7fff10e8c0b0, 
    marshal_data=0x7f7e3a22da1a <gtk_window_focus_in_event>)
    at gtkmarshalers.c:85
#19 0x0000003a78e0faf8 in g_closure_invoke () from /lib64/libgobject-2.0.so.0
#20 0x0000003a78e2119f in signal_emit_unlocked_R ()
   from /lib64/libgobject-2.0.so.0
#21 0x0000003a78e28c34 in g_signal_emit_valist ()
   from /lib64/libgobject-2.0.so.0
#22 0x0000003a78e29702 in g_signal_emit () from /lib64/libgobject-2.0.so.0
#23 0x00007f7e3a20f2f5 in gtk_widget_event_internal (widget=0x22400c0, 
    event=0x20c6530) at gtkwidget.c:6703
#24 0x00007f7e3a20e944 in gtk_widget_event (widget=0x22400c0, event=0x20c6530)
    at gtkwidget.c:6360
#25 0x00007f7e3a060776 in gtk_main_do_event (event=0x20c6530) at gtkmain.c:1649
#26 0x0000003a87c52102 in gdk_event_source_dispatch ()
   from /lib64/libgdk-3.so.0
#27 0x0000003a78247e96 in g_main_context_dispatch ()
   from /lib64/libglib-2.0.so.0
#28 0x0000003a782481e8 in g_main_context_iterate.isra.22 ()
   from /lib64/libglib-2.0.so.0
#29 0x0000003a7824828c in g_main_context_iteration ()
   from /lib64/libglib-2.0.so.0
#30 0x0000003a7969bf34 in g_application_run () from /lib64/libgio-2.0.so.0
#31 0x0000000000422ab0 in main ()

[Peng, 2013-03-17 20:03:38]

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode694
bus/inputcontext.c:694: */
On 2013/03/14 07:00:53, fujiwara wrote:
> On 2013/03/12 18:12:21, Peng wrote:
> > Sorry. I can not understand below code.
> 
> If I press Alt+Tab and move the focus from gedit to thunderbird, the following
> focus events are happened:
> 
> 1) (gedit:445): IBUS-WARNING **: 15:32:36:717  focus_out
> 2) (thunderbird:495): IBUS-WARNING **: 15:32:36:866  focus_in
> 3) (gedit:445): IBUS-WARNING **: 15:32:36:875  focus_in
> 4) (gedit:445): IBUS-WARNING **: 15:32:36:890  focus_out
> 
> 1) It seems XI2 Alt+Tab causes focus-out event at first.
> 2) focused_context is changed from gedit to thunderbird.
> 3) It seems Alt+Tab causes focus-in event when Alt+Tab UI is closed and
> focused_context is changed from thunderbird to gedit.
> 4) It seems Alt+Tab causes focus-out event finally and focused_context is
> changed from gedit to faked_context in ibus global mode.
> 
> Then when _ic_process_key_event() is called, the context in thunderbird does
not
> have the focus and the proxy engine because the faked_context has those.
> 
> This patch applies the proxy engine from faked_context to the context if the
> context->engine is NULL.

I think this comment does not give enough information about this issue. Could
you please add more detail of this issue as comments here?

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode695
bus/inputcontext.c:695: if (context->has_focus && context->engine == NULL &&
Could you please merge this if statement with below if statement?
Like:

if (context->has_focus && context->fake == FALSE) {
  if (context->engine == NULL) {
    ...
  }
  bus_engine_proxy_process_key_event (context->engine, ...);
}

[fujiwara, 2013-03-18 07:04:29]

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/1/bus/inputcontext.c#newcode695
bus/inputcontext.c:695: if (context->has_focus && context->engine == NULL &&
On 2013/03/17 20:03:38, Peng wrote:
> Could you please merge this if statement with below if statement?
> Like:
> 
> if (context->has_focus && context->fake == FALSE) {
>   if (context->engine == NULL) {
>     ...
>   }
>   bus_engine_proxy_process_key_event (context->engine, ...);
> }

I updated comments.
I think there is a case that focused_context->engine is null.
       if (context->engine) {
           bus_engine_proxy_process_key_event (context->engine, ...);
       } else {
       }
I think the if statement is still needed.

[Peng, 2013-03-18 14:37:55]

lgtm with several comments

https://codereview.appspot.com/7725045/diff/8001/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/8001/bus/inputcontext.c#newcode692
bus/inputcontext.c:692: * the following focus events are happened in
ibusimcontext:
events are recevied or events happen

https://codereview.appspot.com/7725045/diff/8001/bus/inputcontext.c#newcode700
bus/inputcontext.c:700: * The problem can be happened without snooper and the
focus in/out
happen

https://codereview.appspot.com/7725045/diff/8001/bus/inputcontext.c#newcode704
bus/inputcontext.c:704: * focused_context here.
Seems it could be put in one line?

[fujiwara, 2013-03-21 02:24:59]

https://codereview.appspot.com/7725045/diff/8001/bus/inputcontext.c
File bus/inputcontext.c (right):

https://codereview.appspot.com/7725045/diff/8001/bus/inputcontext.c#newcode692
bus/inputcontext.c:692: * the following focus events are happened in
ibusimcontext:
On 2013/03/18 14:37:55, Peng wrote:
> events are recevied or events happen 

Done.

[fujiwara, 2013-05-03 16:57:51]

On 2013/03/18 14:37:55, Peng wrote:
> lgtm with several comments

GNOME-Shell side fixed focus issue in Alt+Tab:
https://bugzilla.gnome.org/show_bug.cgi?id=696259

Probably I will close this CL and leave this until we will need this patch for
other issues.

[fujiwara, 2014-09-08 04:38:08]

Finally I decide to integrated this patch to fix a new bug
https://bugzilla.redhat.com/show_bug.cgi?id=1136623 .