Re: [gtk-list] Re: Problem: no key press/release events.

[Tomi Ollila <Tomi Ollila tfi net>]

Monday Aug 16 11:28:48 -0400 1999 Havoc Pennington <hp@redhat.com> wrote:
> 
> On Mon, 16 Aug 1999, Tomi Ollila wrote:
> > 
> > OK. I read the sample manual at: http://pobox.com/~hp/gnome-app-devel.html
> > and noticed the problem is w/ `keyboard' focus. If I understood right, 
> > W/ G[DT]K only toplevel window receives keyboard events and it redirects
> > the data to the widget in question.
> > 
> > I reorganized by code a bit to test this, and indeed -- when I add the
> > keypress/release masks to the toplevel window and connect the signals
> > accorginfly, I get the events... 
> > 
> > Now one thing I yet not know: How to add my eventbox widget to the list
> > of widgets where the toplevel window enters focus?
> > 
> 
> You just have to gtk_widget_add_events() to add KEY_PRESS_MASK to your 
> event box, then set the GTK_CAN_FOCUS flag:
> 
>  GTK_WIDGET_SET_FLAGS(widget, GTK_CAN_FOCUS)

> HTH,
> Havoc

Whoo-oo, at last I got It working. For a long time whatever I did I could
not get keypress/release events to my fixed widget (removed eventbox
since It was unnecessary...) I changed the order of lines in hope that
it would solve the problem...

At last I tried to press <TAB> and then I started getting keypress events.

Now, next saga was to find how to set focus to the widget by default
(Already tried GTK_WIDGET_SET_FLAGS(widget, GTK_HAS_FOCUS) without help).
Browsing through gtkwindow.c I found the solution:

gtk_window_set_focus(GTK_WINDOW(window), fixed).

Next thing was to make button press over the fixed widget to `focus' the
widget unless already focused. What is the "right" way to chech such
a condition. I used the following:
	if (GTK_WINDOW(window)->focus_widget != fixed)...
Does not look as future-compatible as it could... but is it ?


I have one unresolved problem as well..

When pressing key and hold it pressed, many systems starts to generate
a stream of keypress/key release events. The key release event seems to
have the same timestamp as the next key press event...

I wanted the most simple way to regognize that the key is still pressed
in these conditions. First I tried to call `gdk_events_peek()' to peek
the next event... but it would never have anything available. I checked
the code and noticed that It only checks the gdk event queue, but not
from X connection. (Next I tried to call gdk_queue_events(), but that is
static function)... nothing in the gdk interface I explored gave me
the functionality to solve this...

The approach above is hacky, I know...

Does anyone have a good (or even bad) solution how to simply do the feature
I wanted above. I could setup a short timeout or use idle function (and
eventually do if there is no other solutions), but that would complicate
the code further... 

(I know, this won't work on all hardware, but is a advantage on those
 systems that it works).


Tomi