Libretro upstreaming #1464
Libretro upstreaming #1464
Conversation
…ing with the libretro system, but with the existing sdl one (No muting, pausing usw.)
Ghabry
force-pushed
the
libretro_upstreaming
branch
from
c284aab
to
dfd618f
Compare
|
updated the readme with minimal instructions |
Ghabry
force-pushed
the
libretro_upstreaming
branch
from
dfd618f
to
2a2b736
Compare
|
should be ready now. |
Ghabry
force-pushed
the
libretro_upstreaming
branch
3 times, most recently
from
871a0b0
to
864838f
Compare
| #else | ||
| # define Offset off_t | ||
| # define EasyRPG_Offset off_t |
There was a problem hiding this comment.
This is very ugly (also the other type bending in filefinder for PSVita). I hope we can change this to use the same types later and get rid of all these Sce stuff. The vitasdk should really abstract this away in their c library.
There was a problem hiding this comment.
should be done in one go with the AMIGA symbol mess :/
Ghabry
force-pushed
the
libretro_upstreaming
branch
3 times, most recently
from
079663e
to
04f0582
Compare
…(For display refresh > 60 Hz)
…es instead of 735
This is for preventing a crash on Windows, when libretro passes nullptr but CommandLineToArgvW returns the real commandline.
Fix more whitespace issues
…ents all double frees. Minor comment and style changes.
Ghabry
force-pushed
the
libretro_upstreaming
branch
from
d5d71f5
to
534fedc
Compare
carstene1ns
removed
the
Awaiting Rebase
| return false; | ||
| } | ||
|
|
||
| extract_directory(parent_dir, game->path, sizeof(parent_dir)); |
There was a problem hiding this comment.
This crashes on Windows because path in game->path is not UTF-8. Don't tell me they use the local Codepage here m(
There was a problem hiding this comment.
Nevermind, the path is UTF-8 encoded but liblcf can't handle such paths on Windows. In the GameBrowser we work around this with "SetCurrentDirectory" but I have a proper solution for this in my VFS branch, so I won't fix this. This is only a problem for the filename of the main game folder (where liblcf must have access to).
|
This is also ready now. There were no new crashes or anything reported by libretro users. The Player still crashes on Android when closing but I blame RetroArch for this, even my start/stop simplification didn't help here. |
|
I have an idea how to fix the audio thread crashes on core shutdown on Windows and Android (these are RetroArch bugs because they call code in the unloaded dll/so) but I need to wait for there awesome buildbot infrastructure first before I can test it ^^' |
|
okay the workaround works, will only update libretro-common to the latest submodule version, then we are good to go. |
… Keyboard config -> Input, update submodule
Ghabry
force-pushed
the
libretro_upstreaming
branch
from
73e4cd5
to
c09b4f8
Compare
…argc instead, otherwise this mismatches when e.g. libretro uses (nullptr, 0) and crashes. Remove dead code Fix memory leak, the Windows CLI was not deleted
I'm slowly annoyed by always rebasing this therefore I would like to finally upstream the libretro code.
The code is not very invasive, only adds required libretro_ui and _audio files and "USE_LIBRETRO" where needed.
CMake considerations:
The CMakeLists is sufficient to build a easyrpg_libretro.a/.so when you have all dependencies installed (I refuse to add dependency source code in the upstream repo), simply compile with
-DPLAYER_TARGET_PLATFORM="libretro" -DBUILD_SHARED_LIBS=ONand you get a library that works with at least Linux RetroArch.The exception is libretro-common which I added as a submodule with an appropriate CMakeLists.txt.
I'm a bit surprised how simple it was to add a non-SDL2 build with not too many changes (also a nice base now for other platforms) as I didn't intended that use-case when writing the file...
Biggest changes to the CMake file is the new "PLAYER_TARGET_PLATFORM"
Almost everything else is just adding tabulators because code was moved into if() but this renders badly in the diff viewer :/.