Re: [hatari-devel] Regression joystick subsystem

[Matthias Arndt <marndt@xxxxxxxxxxxxxx>]

Am Sun, 14 May 2023 22:06:36 +0200
schrieb Christian Zietz <czietz@xxxxxxx>:

> Matthias Arndt schrieb:
> 
> > When I start Hatari in fullscreen mode with joystick support
> > enabled, e.g. USB gamepad, it seems to work fine and all movements,
> > buttons do work.
> > If I now toggle fullscreen to windowed mode, the joystick stops
> > working. Also reconfiguration to keyboard joystick seems not to
> > work.
> >
> > This is with the SDL2 backend.
> >
> > I can confirm the regression as a Hatari binary build in 2021 from
> > my backup works ok with the same configuration and game image,  
> 
> Could it be (again) a problem with your *local* build instead of
> within Hatari? I just tested the joystick support with a PS4
> controller under Windows running Hatari e8a902a5, using SDL 2. No
> issues occur when switching from fullscreen to windowed mode (and
> back). In other words: I cannot confirm what you're observing.
> 
> Regards
> Christian
> --
> Christian Zietz  -  CHZ-Soft  -  czietz@xxxxxxx
> WWW: https://www.chzsoft.de/
> New GnuPG-Key-ID: 0x8708B34C827B159E
> 
> 
> 

As I told you I am myself not sure anymore. I used fresh directories
with no local changes, unpacking of tarballs from the given changesets
or a fresh git clone (as those were the initial suggestions).

What I do in the CMake settings:

set build type to Release, installation to /opt/hatari and disable
SMALL_MEM

What puzzles me is that the old version freshly compiled has issues
while the old compiled version works, That is very weird and if you
can't reproduce it, I am even more puzzled.

If it cannot be repoduced, that is fine for me. I'd still would like to
understand the problem.

I can check if building with clang instead of gcc makes a difference.

Best regards,
Matthias
-- 
https://final-memory.org/