Hajime Hoshi
7ca3a68eb4
internal/ui: handle an error when navigator.keyboard.getLayoutMap() fails
2025-09-24 22:55:30 +09:00
Hajime Hoshi
8d421df739
internal/ui: bug fix: move handling mod keys to ui.InputState
...
Closes #3320
2025-09-24 04:57:33 +09:00
Hajime Hoshi
7de26d3a30
all: update gomobile
2025-09-23 18:43:39 +09:00
Hajime Hoshi
27b6b3c633
internal/ui: bug fix: key/mouse button inputs were sometimes missed
...
If a key down and a key up happened in one tick, Ebitengine ignored
these events and users failed to notice the key press. This was
problematic in real world use caess, e.g. click by a tap on macOS
browsers.
This change fixes this issue by recording the tick for both a press
and a release, and determine whether the key/button is pressed or
not more precisely.
Updates #2898
Closes #3137
Closes #3249
Updates #3317
2025-09-23 18:04:39 +09:00
Hajime Hoshi
c15408bc6e
internal/ui: update key/mousebutton states at event handlers instead of updateInputStateImpl
...
This is a preparation for more precise key state detections and
asynchronous event handlers.
Updates #1704
Updates #3249
2025-09-23 15:32:19 +09:00
Hajime Hoshi
74835fb5bd
all: add #cgo optimization directives
...
Closes #3196
2025-09-22 12:27:34 +09:00
Hajime Hoshi
5afdb9588f
internal/atlas: replace runtime.SetFinalizer with AddCleanup
...
Closes #3204
2025-09-22 01:52:06 +09:00
Hajime Hoshi
4daa59b2cd
internal/atlas: do not access global variable references at deallocate
...
As deallocate is invoked from the finalizer, it is problematic to access
a global variable referring the object itself. Especially,
runtime.AddCleanup doesn't work well with the current deallocate
implmenentation.
This change removes the access to the global variables from deallocate.
The global variables are reset every frame so it is self not to
remove a deallocated image explicitly. Also, this change fixes
the logic to iterate the global variables' images to check if the
image is already deallocated or not.
Updates #3204
2025-09-22 00:18:20 +09:00
Hajime Hoshi
765193a40e
internal/atlas: refactoring: rename smallImageSet to imageSmallSet and use slices.Delete
2025-09-21 23:51:59 +09:00
Hajime Hoshi
51cc3aed04
all: replace runtime.SetFinalizer with AddCleanup
...
Updates #3204
2025-09-21 23:14:53 +09:00
Hajime Hoshi
244510643c
ebiten: keep TouchID int by using a generics function
2025-09-21 18:48:41 +09:00
Hajime Hoshi
6aa208e15e
ebiten: refactoring: add internal/inputstate
...
internal/inputstate will be accessed from inpututil in the future.
Updates #3249
2025-09-21 18:37:27 +09:00
Hajime Hoshi
97e145a493
all: update Oto to v3.4.0-alpha.11
2025-09-21 01:00:27 +09:00
Hajime Hoshi
77a8c30ca8
internal/file: add VirtualFS.ReadDir and FileEntryFS.ReadDir
...
Closes #3084
2025-09-20 19:24:39 +09:00
Hajime Hoshi
fa51d1d012
examples/dropfile: use slog
2025-09-20 18:56:23 +09:00
Hajime Hoshi
6d949e119f
audio: bug fix: the audio source for testing was too short
...
Closes #3216
2025-09-20 16:41:18 +09:00
Hajime Hoshi
a110d83acd
internal/gamepaddb: update the database
2025-09-20 15:39:25 +09:00
Hajime Hoshi
0459d06eaa
all: update PureGo to v0.9.0-alpha.11
2025-09-19 13:31:15 +09:00
Hajime Hoshi
4943d0e035
internal/ui: cache the monitor
...
`UserInterface.currentMonitor` seems a heavy function and takes about
1% CPU usage on MacBook M3 Pro. Cache the monitor object and use it.
The cache is invalidated after about one second, or the monitor is
missing.
This reduces CPU usages from about 6.5% to 5.5%
Updates #3318
2025-09-19 02:32:28 +09:00
Hajime Hoshi
cc7eac7ba6
all: update Oto
2025-09-18 16:18:42 +09:00
Hajime Hoshi
4bb016f262
internal/processtest/testdata: bug fix: skip issue2737 at GitHub Actions
...
An audio driver seems not found even on Windows.
2025-09-16 16:26:41 +09:00
Hajime Hoshi
9b1f5cbb6a
all: update dependencies and drop Go 1.23
2025-09-16 15:24:16 +09:00
Hajime Hoshi
187c074ca1
vector: skip a flaky test for browsers
2025-09-06 04:06:11 +09:00
Hajime Hoshi
d234fdead6
vector: refactoring: remove unused functions
2025-09-06 03:16:45 +09:00
Hajime Hoshi
159eea683d
vector: calculate the allowance value based on the size of a curve
...
This should improve the result of transforming a path with a
geometry matrix.
Updates #3311
2025-09-06 03:07:21 +09:00
Hajime Hoshi
793b869e73
vector: remove a hack at Close
...
This hack was introduced at 97ebef8298
2025-09-06 02:52:00 +09:00
Hajime Hoshi
be98d0111c
vector: add test for GeoM with arcs
...
The issue was already fixed by fcd3719a9aCloses #3311
2025-09-06 00:58:22 +09:00
Hajime Hoshi
fcd3719a9a
vector: use radius to detect too close two points for ArcTo
...
Using a constant value is not good in the case when the path is
converted with a geometry matrix later.
Updates #3311
2025-09-06 00:45:22 +09:00
Hajime Hoshi
8c7417d07e
vector: path
2025-09-06 00:27:02 +09:00
Hajime Hoshi
329fd15b10
internal/graphicsdriver/metal: use CAMetalDisplayLink for macOS
...
This should fix the flickering issue.
Closes #3278
2025-09-05 00:20:48 +09:00
Hajime Hoshi
95f12febfd
internal/ui: bug fix: crash when the screen size is 0
...
This changea void the crash by skipping to create a screen when
the specified size is 0 accidentally.
Closes #3121
2025-09-04 22:58:31 +09:00
Hajime Hoshi
fd1664adf9
all: update bitmapfont to v4.1.0-alpha.5
2025-08-29 17:23:24 +09:00
Hajime Hoshi
b03fa0532b
internal/cocoa: remove unused APIs
2025-08-28 00:20:42 +09:00
Hajime Hoshi
b99bf14d09
internal/ui: refactoring: rename an argument
2025-08-28 00:19:57 +09:00
Hajime Hoshi
4a0468f29a
internal/graphicsdriver/metal: refactoring
2025-08-27 23:23:23 +09:00
Hajime Hoshi
46fd8a1a52
internal/graphicsdriver/metal: separate command buffers into two
...
For efficiency, this change separatens the current command buffer
into two: for non-screen and for screen. It's because a drawable will
not be released as long as the CommandBuffer referencing it is alive,
it is more efficient to separate CommandBuffers that use the drawable
from those that do not.
Updates #3278
2025-08-27 13:05:21 +09:00
Hajime Hoshi
ecc1def58b
internal/graphicsdriver/metal: refactoring: do not call SetMaximumDrawableCount every frame
2025-08-26 22:39:47 +09:00
Hajime Hoshi
84d4fd45ef
internal/graphicsdriver/metal: fix the case when screenDrawable is nil
...
From the specification, screenDrawable can be nil when present is true.
2025-08-26 19:34:32 +09:00
Hajime Hoshi
a91cf1518f
internal/graphicsdriver/metal: refactoring
2025-08-25 22:50:12 +09:00
Hajime Hoshi
f03b4b9524
internal/graphicsdriver/metal: add assertions
2025-08-25 20:04:38 +09:00
Hajime Hoshi
c830cf4c63
internal/graphicsdriver/metal: do not call PresentDrawable when CommandBuffer is empty
...
Before this fix, PresentDrawable could be invoked even when the
CommandBuffer was empty. This might cause the flickering issue #3278 .
This is an experimental (but seemingly valid) fix.
Updates #3278
2025-08-25 18:39:56 +09:00
Hajime Hoshi
c3a43e057e
internal/graphicsdriver/metal: refactoring
2025-08-25 18:31:11 +09:00
Hajime Hoshi
aa4c2e7267
Revert "internal/graphicsdriver/metal: add in-flight semaphore"
...
This reverts commit db9ca28598#3278
2025-08-25 18:14:55 +09:00
Jon Rafkind
3dc1c8b02b
colorm: extra notes on how colorm operates ( #3307 )
2025-08-25 04:12:28 +09:00
Hajime Hoshi
e18437c213
all: update jakecoffman/cp to v2.3.0
2025-08-23 15:11:03 +09:00
Hajime Hoshi
db9ca28598
internal/graphicsdriver/metal: add in-flight semaphore
...
Closes #3278
2025-08-19 14:08:42 +09:00
Hajime Hoshi
20872c1bd0
cmd/ebitenmobile: fix comments
2025-08-17 12:58:04 +09:00
Hajime Hoshi
a670a741c6
mobile/ebitenmobileview: terminate the process when ebiten.Termination is returned
2025-08-16 00:30:43 +09:00
Hajime Hoshi
e05aadc6ab
internal/vibrate: typo
2025-08-14 17:39:37 +09:00
Hajime Hoshi
371a04f92f
internal/vibration: bug fix: vibration doesn't work when touch feedback was muted
...
Calling `vibrate` without specifying the usage didn't work when the
global setting 'touch feedback' is 0. As the application's vibration
has nothing to do with touch feedbacks, this is an unexpected behavior.
Specify AudioAttributes or VibrationAttributes expcliitly.
Closes #3300
2025-08-14 17:21:02 +09:00
