It shouldn't be RGUI browser dir. Instead it is a more developer centric
option for when libretro apps are distributed (along with RetroArch for
example). Also facilitates a standardized assets folder later on if
applicable.
and get_longitude are gone now in place of get_position. Basically,
from C land we basically do a poll-style queries, but on the
implementation side (ie. Android/iOS/OSX) - they all use callback-based
location updates. So we simply check in the poll function (get_position)
whether position has changed, and if so, update the pointer values and
return true - if not, set them to 0 and return false.
When disabling audio and/or muting, the core has no way
of knowing this, causing potential issues in a threaded mixer.
This environ is not used by any public core to my knowledge so the old
API has been removed and replaced with a change to environment number
(20 -> 22).
retro_run - we need to escape first before we can get the next libretro
core with input data running. This is not needed when we do exec without
being in the main program loop. Anyway, one does not work for the other -
so we need both.
Moved private stuff to libretro_private.h.
Dropped use of retro_variable (redundant and wrong type).
Didn't understand difference between EXEC_LOAD and EXEC_RELOAD at all.
Only one was used anyways ...
to write internal code - other libretro implementations can now just implement
this environ callback and have other external frontends of their own be able to swap
in/out libretro cores from outside
If retroarch was started with g_settings.libretro as a directory, and a
ROM loaded. Going into RGUI would see g_settings.libretro as a directory
and rgui->info wouldn't be properly initialized.
Fixed by updating g_settings.libretro if a specific core is resolved.
Not sure how this will work on RARCH_CONSOLE, but it'll probably be
checked for in startup on core load somehow and either start in menu or args->no_rom.
Allows libretro core to not require a ROM to be loaded.
Kill stdin ROM loading support (pointless these days).
Replace "no ROM on CLI" behavior to be attempting to load a core with no
ROM.
