Currently, the RESOLUTION_FACTOR preference is being set with the GRAPHICS_API key and default. Therefore, it will set/retrieve the wrong values
This revision updates the RESOLUTION_FACTOR preference to use the RESOLUTION_FACTOR key and default value. As a result, RESOLUTION_FACTOR and GRAPHICS_API should store and return the correct (separate) values
* dist: Remove duplicated Finnish translation
For some reason, we had Finnish listed twice on Transifex, causing it be shown twice in Citra.
It has already been deleted again from Transifex, now we only need to remove it from the repo as well.
* citra_qt/configure_ui: Show country of language in the combobox
This prevents an issue where we had seperate versions of the same language for different regions and they were not distinguishable (e.g. "Chinese (China)" and "Chinese (Taiwan)").
* geometry_pipeline: Remove unneeded assert
* Has been hw-tested that gs works correctly even when not in exclusive mode
* pica_core: Propagate output_mask to gs
* Has been hw-tested to occur under the same conditions that other uniforms are shared
* regs_shader: Intialize GPUREG_SH_INPUTBUFFER_CONFIG to default value
* Default value verified on hw. Tales of Abyss does not update the number of vertex attributes for the geometry unit and expects it to be 2
* texture_codec: Align buffer sizes to bpp
* Prevents out of bounds texture reads when launching TOA from the HOME menu
* pica_core: Make default value more clear
The Uniform Read and Address Register Offset ran into precision
differences between f32 and f24, which can be easily fixed and
are not the point of the tests.
As for the LG2 and EX2 tests that were failing, they were wrong.
While it is true that 2^79.7 ~= 1e24, the value is bigger than the
biggest representable value in f24, therefore both EX2 and LG2(?)
should result in Inf.
infinities for all values
doing it at FromFloat32 results in this being applied for all
float<M, E> that aren't constructed from raw.
Note: due to lack of compiler support for C++23 at the moment,
the use of std::isnormal and std::abs results in Trunc not being
constexpr, which required the changes to Zero, One, and operator-,
to prevent FromFloat32 being used in constexpr contexts, and those
specific changes may be reverted in the future.
While I was looking at the NEWS sysmodule, I noticed the params order
for this command were backwards: the info type is the first param,
followed by the buffer size.
This is accurate to my reverse engineered code for the NEWS sysmodule.
Upon further research, I found out the unknown value in FS::Get[This]SaveDataSecureValue indicates that the requesting process is a game card. I have updated the comment for future reference.
* Add random delay to app main thread
* Suggestions
* Remove randomness, only delay with lle
* Apply suggestions
* Fix clang format
* Fix compilation (again)
* Remove unused include
* Implement some missing/wrong AC functionality.
* Schedule NDM connect event into the future
* Disable NDM connect for now as it's causing issues
* Apply latest changes and suggestions.
* Workaround to fake wifi connection.
* Add missing command to ac:i
* Fix compilation
* Fix error codes for CamcelConnectAsync
* Fix missing global state.
* do not move constant variables
* applet_manager: avoid possible use after move
* use constant references where pointed out by msvc
* extra_hid: initialize response
* ValidateSaveState: passing slot separately is not necessary
* common: mark HashCombine as nodiscard
* cityhash: remove use of using namespace std
* Prefix all size_t with std::
done automatically by executing regex replace `([^:0-9a-zA-Z_])size_t([^0-9a-zA-Z_])` -> `$1std::size_t$2`
based on 7d8f115
* shared_memory.cpp: fix log error format
* fix compiling with pch off
