Re: [hatari-devel] Hatari window embedding issue (was: Preparing hatari 2.3.1)

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

Thomas, any ideas about the issue I'm having?

On 12/15/20 12:07 AM, Nicolas Pomarède wrote:
> Le 14/12/2020 à 23:02, Laurent Sallafranque a écrit :
> > I'm working on a music fix to Lockup demo, but it will have to be 
> > integrated in the next release of Hatari.
> >
> > I won't be in time for the 2.3.1
>
> ok, no problem ; at least there's already some changes for falcon in 2.3.1.
> If other changes are more complex, it's better to take the time to test 
> in depth and include them in next major release.

I just fixed couple of minor issues with Python
UI, but while testing Hatari hard disk handling
changes I noticed rather bad issue with Python
Gtk UI embedding.

While everything works fine with VDI mode and
ST/STE screen handling, Hatari shows just black
screen when user starts with, or switches to
either TT or Falcon emulation (= videl.c based
render mode).


Resizing the Python UI window doesn't help.
"screen.c" debug output shows Hatari calculating
new scale factors so it gets the resize events,
but screen still keeps black.

Only by switching Hatari to fullscreen and back
gets Hatari to start showing its screen in Python
UI.

I've read screen.c code, but I don't see where
things go wrong.  All the code paths should be
calling Control_ReparentWindow() to reparent
Hatari window, and as resize events cause Hatari
to act, I think that should work.

One problem could be Hatari not getting expose
event(s) after its window is reparented, but I
don't understand why that would work differently
between ST/STE and TT/Falcon mode.

In case it's more of a SDL2 than Hatari issue,
I would be interested if others can reproduce
the issue with newer SDL2 version than I have
(v2.0.9).  Or different Gtk version (v3.22).


	- Eero