Commit Graph

950 Commits

Author SHA1 Message Date
Hajime Hoshi
58d3655597 internal/ui: remove unnecessary receivers from methods 2023-09-24 19:45:55 +09:00
Hajime Hoshi
b32575b7b2 internal/ui: refactoring: add (*Monitor).deviceScaleFactor 2023-09-24 19:41:37 +09:00
Hajime Hoshi
0e23045b90 internal/ui: refactoring: add (*monitors).monitorFromPosition 2023-09-24 19:41:28 +09:00
Hajime Hoshi
fab08bac92 internal/ui: use RLock/RUnlock whenever possible 2023-09-24 19:27:22 +09:00
Hajime Hoshi
4fb27adb9d internal/ui: remove an old comment
Now Ebitengine allows to specify the initial monitor, having an
initial monitor is no longer a hack.

Updates #1575
Updates #1835
2023-09-24 19:24:23 +09:00
Hajime Hoshi
b5d0dc3686 internal/ui: change the initial window position adjustment
The window position should be posible if possible.
2023-09-24 19:07:21 +09:00
Hajime Hoshi
a23efac01c internal/ui: refactoring: reduce member variables 2023-09-24 19:05:56 +09:00
Hajime Hoshi
c16bd1e249 internal/ui: refactoring: unify duplicated logics to initialize the window size and position 2023-09-24 18:55:45 +09:00
Hajime Hoshi
c7d1d28582 internal/ui: move impls for device scale to internal/ui for desktops
This enables to remove restrictions for some functions to be called
from the main thread.

Updates #2423
2023-09-24 15:44:41 +09:00
Hajime Hoshi
458a415131 internal/ui: move impls for device scale to internal/ui for mobiles 2023-09-24 15:29:14 +09:00
Hajime Hoshi
506a1de259 internal/ui: refactoring: move device-scale implementation to internal/ui for browsers
The returned value from internal/devicescale.At never changes for
browsers, so the detection of devicePixelRatio updates didn't work
in the first place. Also, there is not a good way to detect the
change [1].

This change moves the logic from internal/devicescale to internal/ui.
We aim to merge these packages as a device scale factor belongs to
a monitor and internal/ui manages monitors.

[1] https://crbug.com/123694
2023-09-24 03:19:31 +09:00
Hajime Hoshi
5d5e3c7c0b internal/ui: refactoring: replace *glfw.Monitor with *Monitor 2023-09-24 02:39:53 +09:00
Hajime Hoshi
f72b8a4ced internal/ui: refactoring: remove redundant Monitor members 2023-09-24 01:58:07 +09:00
Hajime Hoshi
121d6005cd internal/ui: remove (*Monitor).Bounds
Updates #2780
2023-09-24 01:46:14 +09:00
Hajime Hoshi
01e2c92e56 internal/ui: remove unnecessary code
This was an old code when Ebitengine created a dummy invisible
window before creating the main window.
2023-09-24 01:42:39 +09:00
Hajime Hoshi
69e3a2b974 internal/ui: refactoring: (*monitors).update must be called from the main thread
In the current implementation, (*monitors).update is not called
from other threads, but the current code is fragile.
2023-09-24 01:37:11 +09:00
Hajime Hoshi
7664647ad1 internal/ui: bug fix: needed to wait before entering into fullscreen
Just after moving a window to another monitor, waiting a little while
was required before entering into fullscreen.

Closes #2778
2023-09-23 19:02:36 +09:00
Hajime Hoshi
4ef98b3ea9 internal/ui: bug fix: wrong calculation of the initial window position
Updates #2778
2023-09-23 18:51:58 +09:00
Hajime Hoshi
b3058b68a0 ebiten: change the unit of (*Monitor).Bounds
Monitors can have different device scale factors, and in this case,
it doesn't make sense to use device-independent pixels as unit for
monitor positions and sizes.

Updates #2778
2023-09-23 17:58:23 +09:00
Hajime Hoshi
d736a8cff7 internal/ui: bug fix: GetVideoMode must not be called from othere threads
Now setInitMonitor can be called from any other threads than the main
thread. Do not call this.

This change fixes the issue by using *Monitor instead of *glfw.Monitor.
2023-09-23 17:30:18 +09:00
Hajime Hoshi
e0cd031aea internal/ui: bug fix: replace initWindowMonitor usages with initMonitor
There were two similar member variables internalWindowMonitor and
initMonitor, and when SetMonitor is called, probably there were
some inconsitency between them.

This change fixes the issue by not using initWindowMonitor.

Closes #2779
2023-09-23 17:17:38 +09:00
Hajime Hoshi
14f2ee198e ebiten: add SetWindowMousePassthrough and IsWindowMousePassthrough
Closes #2511
2023-09-18 18:31:22 +09:00
Hajime Hoshi
8a0aba45f5 internal/ui: update touches after the layout is determined on mobiles
Closes #2763
2023-09-17 21:56:26 +09:00
Hajime Hoshi
02fd8cfb07 internal/ui: bug fix: preserve a captured cursor positions for toggling fullscreening for browsers
Closes #2475
2023-09-17 18:38:00 +09:00
Hajime Hoshi
a5aa721bda internal/ui: update mouse and touch cursors after the layout is determined on browsers
Updates #2763
2023-09-17 16:30:47 +09:00
Hajime Hoshi
6ab509f221 internal/ui: reland: bug fix: preserve cursor positions for toggling fullscreening for desktops
Updates #2475
2023-09-17 15:54:39 +09:00
Hajime Hoshi
9d73e45677 Revert "internal/ui: bug fix: preserve cursor positions for toggling fullscreening for desktops"
This reverts commit 7ed4db90be.

Reason: compile error on Windows
2023-09-17 15:43:38 +09:00
Hajime Hoshi
7ed4db90be internal/ui: bug fix: preserve cursor positions for toggling fullscreening for desktops
Updates #2475
2023-09-17 15:26:08 +09:00
Hajime Hoshi
7becaa19e6 internal/ui: bug fix: native APIs must be called from the main thread
Updates #2763
2023-09-17 15:14:57 +09:00
Hajime Hoshi
0475baf9e2 internal/ui: update comments
Updates #2763
2023-09-17 14:59:43 +09:00
Hajime Hoshi
8c25b07336 internal/ui: call updateInputState after layoutGame
The cursor position is affected by the current layout. Then, input
states should be updated after layoutGame is called.

Updates #2763
2023-09-17 14:58:32 +09:00
Hajime Hoshi
a62b8a00e7 internal/ui: use float64 for cursor positions internally 2023-09-17 03:43:18 +09:00
Hajime Hoshi
45acc2c5d8 internal/ui: reland: bug fix: disable window size limitation on fullscreen
Closes #2260
2023-09-16 16:49:49 +09:00
Hajime Hoshi
40fc96d78c internal/ui: forbid a green button when a max window size limit is specified
Updates #2260
2023-09-16 16:40:03 +09:00
Hajime Hoshi
128215eedd Revert "internal/ui: bug fix: disable window size limitation on fullscreen"
This reverts commit 9a5ef1b553.

Reason: we should disable the green button instead

Updates #2260
2023-09-16 03:23:34 +09:00
Hajime Hoshi
657e4c2541 internal/ui: update a comment
Closes #2240
2023-09-16 03:12:37 +09:00
Hajime Hoshi
619436a47b internal/ui: refactoring: add monitor_glfw.go 2023-09-16 00:16:00 +09:00
Hajime Hoshi
dd2768d5f3 internal/ui: bug fix: introduce locks for monitors
Updates #1853
2023-09-15 03:59:13 +09:00
Hajime Hoshi
32bd565df9 internal/ui: bug fix: setWindowMonitor was processed even when the same monitor was specified
Updates #1835
2023-09-15 03:30:15 +09:00
Hajime Hoshi
b710ff17c8 internal/ui: bug fix: wait for a while after exiting fullscreen on macOS
Closes #2758
2023-09-15 03:19:37 +09:00
Hajime Hoshi
90accfba1e internal/ui: bug fix: do not call PollEvents on Windows and Linux
Updates #2296
2023-09-15 03:11:58 +09:00
Hajime Hoshi
568e98ef1c internal/ui: bug fix: call setOrigWindowPos on entering fullscreen
Closes #2757
2023-09-15 02:53:38 +09:00
Hajime Hoshi
f7a801d591 internal/ui: refactoring 2023-09-15 02:45:31 +09:00
Hajime Hoshi
9a5ef1b553 internal/ui: bug fix: disable window size limitation on fullscreen
Closes #2260
2023-09-15 02:19:20 +09:00
Hajime Hoshi
41c7852c7d internal/ui: sort selectors in alphabetical order 2023-09-11 02:15:42 +09:00
TotallyGamerJet
5dc293fe4f
internal/ui: use new OBJC API (#2750)
Updates #1162
2023-09-11 02:05:42 +09:00
Hajime Hoshi
f30a58a393 internal/ui: bug fix: disable global functions after the game termination
Closes #2743
2023-09-02 15:38:41 +09:00
Hajime Hoshi
6e968558b1 internal/ui: bug fix: setRunning(true) must be called after the main thread is set
Some functions like ebiten.SetCursorMode use `running` to detect
whether the game starts or not. If the game starts, the main thread
must exist, but there was a timing when `running` was true but the
main thread didn't exist.

This change fixes this issue by changing the timing to call
`setRunning(true)` after the main thread initialization and before
`initOnMainThread`. `initOnMainThread` assumes that `running` is
true.

Closes #2742
2023-09-02 04:24:51 +09:00
Ketchetwahmeegwun T. Southall
60b7de6a3c
ebiten: add APIs to treat monitors (#2597)
This change adds these APIs:

* `type MonitorType`
* `func (*MonitorType) Bounds() image.Rectangle`
* `func (*MonitorType) Name() string`
* `func Monitor() *MonitorType`
* `func SetMonitor(*MonitorType)`
* `func AppendMonitors([]*MonitorType) []*MonitorType`

Closes #1835
2023-08-30 21:02:04 +09:00
Hajime Hoshi
534370f7b1 internal/graphics: enable to specify regions for each source image
This is a preparation to specify different sizes of source images.

Updates #1870
2023-08-27 21:15:01 +09:00
Hajime Hoshi
a9b2f5f9ca Revert "ebiten: panic if a non-existent uniform variable name is given"
This reverts commit 1b8580fab7.

Reason: some existing applications don't work with this fix.

Updates #2710
2023-08-24 14:15:19 +09:00
Hajime Hoshi
f81dbd9288 internal/graphicsdriver: flush commands asynchronously whenever possible
Closes #2664
2023-07-30 22:44:13 +09:00
Hajime Hoshi
5a64f8299e internal/thread: refactoring: add a common interface Thread 2023-07-30 22:42:24 +09:00
Hajime Hoshi
d13bea29dc internal/ui: refactoring 2023-07-30 03:52:54 +09:00
Hajime Hoshi
3869e2e4f6 internal/ui: refactoring: integrate the render thread usages into internal/graphicscommand
Updates #2664
2023-07-30 03:50:49 +09:00
Hajime Hoshi
1b8580fab7 ebiten: panic if a non-existent uniform variable name is given
Closes #2710
2023-07-29 20:24:42 +09:00
Hajime Hoshi
0b1c7404d5 ebiten: use zero values for an unspecified uniform variable
Closes #2709
2023-07-29 19:11:49 +09:00
Hajime Hoshi
d0e4023d88 internal/ui: add a length check for uniform variables 2023-07-29 18:59:36 +09:00
Hajime Hoshi
8b4c744ca8 internal/ui: rename a file
Use `_linbsd.go` suffix for Linux and BSDs.
2023-07-23 02:29:57 +09:00
Hajime Hoshi
91e1c0ea29 ebiten: add more cursor shapes
This change adds these new cursor shapes:

* CursorShapeNESWResize
* CursorShapeNWSEResize
* CursorShapeMove
* CursorShapeNotAllowed

Closes #2476
2023-07-23 00:49:45 +09:00
Hajime Hoshi
7a7d4fd91f internal/ui: bug fix: need to delay to capture a cursor
From the spec https://w3c.github.io/pointerlock/#extensions-to-the-element-interface,
capturing a cursor might require a cooltime. This change adds a delay
to capture a cursor just after escaping from a capture.

Closes #2693
2023-07-07 01:24:02 +09:00
Hajime Hoshi
eb5c1ee818 internal/ui: bug fix: the original cursor position was not scaled correctly
Updates #2690
2023-07-05 09:24:29 +09:00
Hajime Hoshi
929539b66e add exp/textinput package
This works only for macOS and browsers so far.

Updates #1029
2023-07-05 02:09:04 +09:00
Hajime Hoshi
1d02d20879 internal/ui: fix comments 2023-07-04 23:34:00 +09:00
divVerent
b96aea70f1
internal/ui: compute antialias region size correctly. (#2681)
Closes #2679
2023-06-24 02:25:19 +09:00
Hajime Hoshi
0785502be3 internal/ui: refactoring 2023-05-09 00:35:54 +09:00
Hajime Hoshi
d2c991b774 all: refactoring: use image.Rectangle 2023-04-28 01:03:38 +09:00
Hajime Hoshi
00e45affe9 all: update OpenGL version from 2.1 to 3.2
'texelFetch' requires OpenGLSL 1.30, which requires OpenGL 3.0+.
macOS might not support OpenGL 3.0 and 3.1, so adopt 3.2.

Updates #1431
2023-04-23 14:25:36 +09:00
Hajime Hoshi
b40c5b1e99 internal/ui: catch an error at At at updateIconIfNeeded
Closes #2647
2023-04-20 00:59:14 +09:00
Hajime Hoshi
c00579509a internal/ui: bug fix: wrong mouse cursor movement calculation
Closes #2634
2023-04-10 15:25:04 +09:00
Hajime Hoshi
09d2459f1c internal/ui: a more robust fix for the unfocused-window loop
Depending on the member variable bufferOnceSwapped is a little fragile.
Detect whether the window is shown or not instead.

Updates #2620
2023-04-01 03:36:51 +09:00
Hajime Hoshi
db7c99fde8 internal/ui: bug fix: an unfocused window should not wait until buffers are swapped once
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
2023-04-01 03:01:59 +09:00
Hajime Hoshi
14b34a7b0a internal/ui: typo 2023-03-31 03:21:48 +09:00
Hajime Hoshi
1999f8a024 internal/ui: use DirectX as a fallback for old Windows
Initializing OpenGL can fail [1], though this is pretty rare. For such
this case, try DirectX as a fallback.

[1] https://answers.microsoft.com/en-us/windows/forum/all/my-drivers-dont-support-opengl/a1ee0267-4dc8-4385-a185-6ed322d78329

Updates #2613
2023-03-31 03:15:52 +09:00
Hajime Hoshi
370b7599bc internal/graphicsdriver/directx: use OpenGL as the default on old Windows
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
2023-03-31 02:21:11 +09:00
Hajime Hoshi
31273c875f internal/graphics: rename constants 2023-03-23 20:48:39 +09:00
Hajime Hoshi
c7214ef6fb internal/ui: use MaxVertexCount instead of IndicesCount
Updates #2460
2023-03-23 19:44:00 +09:00
Hajime Hoshi
1656c47abf internal/ui: bug fix: hiding a window decoration at initialization doesn't work on macOS
Closes #2600
2023-03-14 00:27:57 +09:00
Hajime Hoshi
e93bcd9e55 internal/ui: bug fix: needed to wait for swapping buffers before entering fullscreen
Closes #2599
2023-03-13 21:40:40 +09:00
Hajime Hoshi
051b0c7238 internal/ui: bug fix: a big-offscreen size could be 0
Closes #2589
2023-03-03 22:34:43 +09:00
Hajime Hoshi
ea7d007df9 internal/ui: bug fix: dstRegion should not exceed the image region
Closes #2577
2023-02-21 14:15:39 +09:00
Hajime Hoshi
d91b8cae88 internal/ui: reland: refactoring: improve anti-alias logic
* Remove the image extension. The image is always flushed when necessary.
* Use a small offscreen image whichever the blend is.

Updates #2399
2023-02-19 23:49:32 +09:00
Hajime Hoshi
3aa8448511 Revert "internal/ui: refactoring: improve anti-alias logic"
This reverts commit 41f8295611.

Reason: test failures: https://github.com/hajimehoshi/ebiten/actions/runs/4216421788/jobs/7318236488
2023-02-19 22:46:37 +09:00
Hajime Hoshi
41f8295611 internal/ui: refactoring: improve anti-alias logic
* Remove the image extension. The image is always flushed when necessary.
* Use a small offscreen image whichever the blend is.

Closes #2399
2023-02-19 22:14:23 +09:00
Hajime Hoshi
961c34efb5 internal/ui: use a small buffer image for anti-alias
Closes #2399
2023-02-19 17:35:43 +09:00
Hajime Hoshi
8479fee606 internal/ui: refactoring 2023-02-18 18:23:57 +09:00
Hajime Hoshi
56597800dd internal/ui: refactoring: add bigOffscreenImage
Updates #2399
2023-02-18 02:32:41 +09:00
Hajime Hoshi
7315044dcf internal/ui: remove a redundant call to close a window
glfw.Terminate should close the window too. This might fix the issue
so let's wait and see.

Updates #2551
2023-02-15 14:36:28 +09:00
Hajime Hoshi
8864ce1291 all: update comments 2023-02-07 12:05:43 +09:00
Pierre Curto
4de807cc44
all: fix typos (#2558)
Signed-off-by: Pierre Curto <pierre.curto@gmail.com>
2023-01-28 19:06:38 +09:00
Hajime Hoshi
f009dd8dd2 ebiten: implement DroppedFiles to replace AppendDroppedFiles
Closes #2552
2023-01-25 14:27:11 +09:00
Hajime Hoshi
16e49e0499 internal/ui: fix a potential out-of-range error 2023-01-24 00:32:08 +09:00
Hajime Hoshi
84e32c2e4e internal/ui: more precise detection of directory on browsers
Updates #1868
2023-01-22 18:28:34 +09:00
Hajime Hoshi
8c25fac860 ebiten: add AppendDroppedFiles
Closes #1868
2023-01-22 17:22:25 +09:00
Hajime Hoshi
b79f0394cc internal/ui: refactoring: allow slices in InputState 2023-01-22 01:28:57 +09:00
Hajime Hoshi
d53803615a internal/ui: merge a window-closing state into an input state 2023-01-21 23:42:48 +09:00
Hajime Hoshi
08fba564f8 internal/ui: refactoring: define global sel_ objects 2023-01-09 01:41:13 +09:00
Hajime Hoshi
95f1ef0fb9 internal/ui: bug fix: recover the cursor shape when the cursor is visible
Closes #2527
2023-01-07 20:03:21 +09:00
Hajime Hoshi
86fa9366f7 internal/ui: use 'set' instead of 'push' for cursors
We don't need to use a cursor stack.
2023-01-07 19:16:41 +09:00
Hajime Hoshi
7418576c16 internal/ui: re-enable skipping to render the final screen when possible
With Metal, nextDrawable could return immediately when a command buffer
is empty. To avoid high CPU usage, this change adds a slight sleep in
this case.

With DirectX, Present waits for a while even though nothing is updated,
then that's fine.

Updates #2341
Updates #2342
Updates #2520
2023-01-03 23:09:28 +09:00