This is basically a revert of 0035ba0bd1.
I couldn't confirm that allowing tearing improved FPS. Rather, this
has a potential risk to cause an error at Present of the swap chain due
to an inconsistent argument, though I have not confirmed an actual issue.
Now the DirectX 11 driver was introduced, the situation has changed.
Updates #2034
Updates #2188
This is achieved by replacing the function by one that only adds a single
new region, and only considers duplicates between the previously existing
region and the one newly added one, thereby removing previously redundant
checking of each previously existing region against each other.
This speeds up AAAAXY loading on a Moto G7 Play from 52.27 seconds to
8.15 seconds.
Closes#2626
By removing redundant work done in canAlloc and just calling alloc right away,
this removes 35% contribution to AAAAXY loading CPU time on the profile on
https://user-images.githubusercontent.com/251568/230496805-c5e32c19-9258-49c8-800b-a3f0bc3b072d.svg,
or - as measured via stopwatch - brings loading time on Moto G7 Play from
17.22s to 14.94s after already applying #2627.
This should be safe as there is no case in which alloc succeeds and the
allocated region isn't used; also, there is no case in which alloc mutates the
tree when it doesn't actually succeed (comment added in one place to justify
this).
Closes#2628
On macOS, a window is hidden until buffers are swapped once. If a game is
not runnable on unfocused, as the window is hidden and unfocused, the
waiting loop for a window never ends.
This change fixes the issue by changing the loop condition.
Closes#2620
On old machines, CreateSwapChain might fail without a specific reasons.
Probably this is due to an issue in drivers, and if so, there is nothing
we can do. When choosing a graphics library, prefer OpenGL on such old
machines.
Updates #2613
CreateDXGIFactory2 is available in Windows 8.1 and newer. This is not
necessary for the current Ebitengine implementation. Use the old version
CreateDXGIFactory to support Windows 8 and older.
Updates #2613
A screen image in DirectX 12 should not include any buffers to release
so there is not a bug, but disposing it explicitly when resizing the
window should be polite.
This change is reland of 54e2790a06
but with a fix for WebGL. Also, this changes the logic when len(vertices)
exceeds the maximum: just remove the last part.
Updates #2460
This commit mirrors https://github.com/libsdl-org/SDL/pull/7405
(6f1f586086).
Note that for SDL, this code is used a lot less than for Ebitengine, as SDL
mostly migrated to HIDAPI and direct USB device access rather than using
Android's APIs. For Bluetooth devices, however, the Java APIs are used the
same way.
This was the remaining problem to be solved to automatically support standard
layout on most standard gamepads (this should cover most Xbox-ish and PS-ish
gamepads on the market).
In particular this covers gamepads with the following assignment:
- Left stick = X/Y, right stick = Z/RZ, triggers = LEFTTRIGGER/RIGHTTRIGGER
(which basically is what Android docs say and some PS gamepads do)
- Left stick = X/Y, right stick = RX/RY, triggers = Z/RZ
(Xbox gamepad style, apparently)
- Left stick = X/Y, right stick = RX/RY, triggers = LEFTTRIGGER/RIGHTTRIGGER
(Not sure if this exists, but it's conceivable)
As we found on the SDL pull request discussion, gamepads that offer flight
controls (e.g. THROTTLE and RUDDER) will likely not work well, before and
after this change.
Closes#2557
pixelsForRestore was removed at 09e0320309
as the regions for restoring were minimized. However, this caused a
regression and increased allocations on Android.
This fix revives pixelsForRestore, but as a map with region keys and
byte slice values. pixelsForRestore no longer represents a byte slice
for an entire image.
Updates #2375
Implements native standard layout for Linux gamepads by using the kernel-provided button IDs, thereby expanding
support to gamepads not listed in gamecontrollerdb.txt.
Linux's docs: https://www.kernel.org/doc/Documentation/input/gamepad.txt
SDL2's source: https://fossies.org/linux/SDL2/src/joystick/linux/SDL_sysjoystick.c#l_1740
Note that I am NOT 100% convinced about the X/Y swap between Xbox and PlayStation controllers - the Xbox
compatible pad I have however does have BTN_NORTH and BTN_WEST swapped (and thus BTN_X and BTN_Y
assigned right), which confirms SDL's logic and opposes the kernel docs.
Tested with this gamepad: "20d6:2802 BDA Xbox ONE Core controller", label says "PowerA Model 1508491-02" - even
after clearing out gamecontrollerdb.txt, examples/gamepad shows a 100% correct mapping.
Closes#2052
