You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Unlike SDL, the headless testing environment doesn't attempt to set the config's default values as specified in the ConfigSetting constructors, resulting in tests failing if the display option isn't set to 60.
What should happen
We need the testing environment to run the code to assign these default values (and only these, not the user config, which may break tests that assume default behaviour).
As far as I can tell, it's the only option causing breakage to the tests.
But who knows what'll happen once we read other potentially non-zero values from g_Config when tests relied on everything being zeroed-out thus far.
In any case, it sounds like an oddity that should be addressed.
Once this is done, this would allow us to move the compile-time constant back to the ConfigSetting constructor if you find it more aesthetically pleasing :p
Game or games this happens in
Unit tests :)
What area of the game / PPSSPP
When adding a display-related option, I had to move its default value assignment to a compile-time constant to please the tests.
Because it was found to be a non-standard way to add defaults to options, I dug deeper to the root cause:
Unlike SDL, the headless testing environment doesn't attempt to set the config's default values as specified in the
ConfigSetting
constructors, resulting in tests failing if the display option isn't set to 60.What should happen
We need the testing environment to run the code to assign these default values (and only these, not the user config, which may break tests that assume default behaviour).
As far as I can tell, it's the only option causing breakage to the tests.
But who knows what'll happen once we read other potentially non-zero values from
g_Config
when tests relied on everything being zeroed-out thus far.In any case, it sounds like an oddity that should be addressed.
Once this is done, this would allow us to move the compile-time constant back to the
ConfigSetting
constructor if you find it more aesthetically pleasing :pLogs
No response
Platform
Linux / BSD
Mobile device model or graphics card (GPU)
Headless
PPSSPP version affected
master
Last working version
No response
Graphics backend (3D API)
Vulkan
Checklist
The text was updated successfully, but these errors were encountered: