[Mono-osx] Mono.framework 1.2.6_6 gtk-sharp issues

[Geoff Norton]

Bob,


First and foremost, thanks for the feedback.

On 11-Feb-08, at 12:00 AM, Bob wrote:

> Note that I posted this on the Imendio GTK+ Mac OS X forum as well.
>
> I have tried the new GTK Mac backend contained in the Mono framework  
> MonoFramework-1.2.6_6.macos10.novell.universal.dmg which was  
> recently released.  I used in on an application I wrote and am very  
> impressed that most things work.  Nice job getting this up to  
> speed.  This is close to being usable.  I am on a PowerBook G4  
> running Mac OS 10.4.10.
>
> I do have a few issues.  Although I am using mono and gtk-sharp, I  
> suspect that the problems I am having are in the GTK backend.  The  
> 1st of these is most important to me, the others are either  
> annoyances or can be worked around.  But #1 completely changes the  
> behavior of my app.
>
> 1. KeyPressEvents seem to be handled differently using the X11  
> backend and the Mac backend.  Using X11 ALL keypress events are  
> queued and handled by the main loop when it gets around to them.  So  
> if I have a big computation going and hit two key presses, the main  
> loop will see them when the big computation stops and the main loop  
> runs.  But with the Mac backend key presses are not be seen by the  
> main loop if they are pressed while a big computation is running.   
> They just disappear.  This greatly changes the behavior of my  
> application (making it very hard to use reliable).
>
> 2. I use a Gtk.DrawingArea. It renders the window using pixmaps (in  
> the background) and then the pixmap is drawn to the Gdk.Window (of  
> the drawing area) using Gdk.Window.DrawDrawable.  The problem is  
> that using the Mac GTK backend this does not cause the actual window  
> to get re-drawn (except when the window is first created).  If I put  
> in an explicit call to Gdk.Window.InvalidateRect after the call  
> DrawDrawable then it gets re-drawn but this should not be necessary  
> (and is not necessary using the X11 backend).
>
> 3. The keypress events for the function keys do not get recognized.   
> When I print them using Console.WriteLine ("KPress: {0}", evnt.Key);  
> they show as "KPress: 16777232" (instead of KPress: F2, etc).  I am  
> using a PowerBook G4 and have to push the "fn" key while pressing  
> the function key so this could be playing a role (but it works fine  
> in X11 mode).
>
> 4. I use fullscreen mode (Gtk.Window.Fullsceeen) and then Hide the  
> fullscreen window when I want to go back to my normal window.  This  
> works well except that the Mac Menu Bar is not re-drawn after the  
> Hide call.  I should mention that I am impressed that the Fullscreen  
> really uses the entire screen (including covering up the dock and  
> mac menu bar).  This is the way it should be and I was afraid that  
> it would not work this was (this is a step up from the X11 backend  
> on Mac).
>
> 5. I use a Gtk.TreeView and the vertical scrollbar sliders are not  
> drawn (the horizontal ones are drawn).  The vertical scrollbar seems  
> to work fine but I have to guess where the slider is. I noticed that  
> the gtk-demo program has this problem also.
>
> 6. The emulated mouse buttons do not work.  For example, control  
> click is suppose to be like a right mouse button (button 2) but  
> instead it shows as a control and then a click.
>
> 7. I get a bunch of "CGBitmapContextGetBitsPerPixel: invalid  
> context" getting output.  Not sure if this matters.  I also noticed  
> that the gtk-demo program included in the distribution spits these  
> out as well.

Could you please file issues for each of these independently as bugs  
and assign them to me (gnorton at novell.com).  The bulk of them are  
likely gtk+ bugs, but we're trying to triage for imendio and make  
native testcases for them as much as we can (and contribute back) and  
this would be very helpful for us.

Thanks!

-g