* Shader compilation is now cached. Subsequent loads take less than a millisecond.
* Improved game, editor and project manager startup time.
* Editor uses .godot/shader_cache to store shaders.
* Game uses user://shader_cache
* Project manager uses $config_dir/shader_cache
* Options to tweak shader caching in project settings.
* Editor path configuration moved from EditorSettings to new class, EditorPaths, so it can be available early on (before shaders are compiled).
* Reworked ShaderCompilerRD to ensure deterministic shader code creation (else shader may change and cache will be invalidated).
* Added shader compression with SMOLV: https://github.com/aras-p/smol-v
- Tweak the setting property hint to be more informative.
- Make the setting a "basic" setting so it appears when Advanced Settings
is disabled.
- Remove redundant orientation setting in the iOS export preset.
The project setting is now used (like on Android).
Projects upgrading from a previous version will have to set the
screen orientation again in the Project Settings if it wasn't set
to the default value ("landscape").
The XDG Base Directory specification does not allow using relative paths
(which broke things in Godot anyway). If a relative path is detected,
it should be ignored.
We found that this flag causes this error on PR #48812 which does not add any
fancy inline assembly:
```
/tmp/tile_set-ce236a.s: Assembler messages:
/tmp/tile_set-ce236a.s:34676: Error: selected processor does not support `bfc x0,#32,#32'
clang++: error: assembler command failed with exit code 1 (use -v to see invocation)
```
That flag is mentioned in various errors related to assembler failures on
arm64v8 with Clang from the Android NDK.
It was added in Godot in #6958 when migrating from GCC to Clang, and is indeed
referenced in the NDK's Clang migration guide:
https://android.googlesource.com/platform/ndk/+/master/docs/ClangMigration.md
> Especially for ARM and ARM64, Clang is much stricter about assembler rules
> than GCC/GAS. Use `-fno-integrated-as` if Clang reports errors in inline
> assembly or assembly files that you don't wish to modernize.
We don't get those errors nowadays so it seems the flag is no longer needed.
Add `WARN_ON_UNDEFINED_SYMBOLS=0` for the main module (which defines
`godot_js_main` as extern coming from the "side" module, i.e. the main
Godot binary).
This changes the types of a big number of variables.
General rules:
- Using `uint64_t` in general. We also considered `int64_t` but eventually
settled on keeping it unsigned, which is also closer to what one would expect
with `size_t`/`off_t`.
- We only keep `int64_t` for `seek_end` (takes a negative offset from the end)
and for the `Variant` bindings, since `Variant::INT` is `int64_t`. This means
we only need to guard against passing negative values in `core_bind.cpp`.
- Using `uint32_t` integers for concepts not needing such a huge range, like
pages, blocks, etc.
In addition:
- Improve usage of integer types in some related places; namely, `DirAccess`,
core binds.
Note:
- On Windows, `_ftelli64` reports invalid values when using 32-bit MinGW with
version < 8.0. This was an upstream bug fixed in 8.0. It breaks support for
big files on 32-bit Windows builds made with that toolchain. We might add a
workaround.
Fixes#44363.
Fixesgodotengine/godot-proposals#400.
Co-authored-by: Rémi Verschelde <rverschelde@gmail.com>
We don't get updates when the window is unfocused/minimized, so we must
detect the situation where the counted ticks start drifting away
resulting in more frames drawn than needed.
This commit adds a check to ensure that the target ticks do not drift
away more than one second.
See discussion in #43811, it was only implemented on iOS and even that
implementation was fairly limited. This would best be provided as plugins
for Android and iOS without cluttering the shared OS API.
We've been using standard C library functions `memcpy`/`memset` for these since
2016 with 67f65f6639.
There was still the possibility for third-party platform ports to override the
definitions with a custom header, but this doesn't seem useful anymore.
Generates a key/cert snakeoil pair or use a custom SSL cert/key.
This is of course false security, and potentially detrimental for it.
But, so long, those are the requirements browser vendors agreed on to
use things like the Gamepad API, and more advanced topics like wasm
threads.
You don't need this if you run on localhost (at least!), but you do
need this (or a much safer nginx proxy) to try those things on your
local network (e.g. when debugging a phone, networking, etc).
We used to only generate the favicon if it was specified in the user
project settings, now it's optional, will export it to `NAME.icon.png`,
(falling back to the default project icon if none is set in project
settings), and the `<link>` tag is added using the `$HEAD_INCLUDE`
instead of being hardcoded in the template.
After further testing it seems to work fine now when building binaries with GCC 5
on Ubuntu 16.04 (previously we were using GCC 9 on Ubuntu 14.04).
Follow-up to #45629.
in uwp's version of export.cpp create a ExportNotifier object so that EditorExportPlugin _export_begin and _export_end functions will be called as documentated.
We were using `Content-Length` from the server when `Content-Encoding`
was not set (i.e. response was not compressed).
Sadly, in CORS requests accessing headers is restricted, and while
`Content-Length` is enabled by default, `Content-Encoding` is not.
This results in the impossibility of knowing if the content was
compressed, unless the server explicitly enabled the encoding header
via `Access-Control-Expose-Headers`.
To keep maximum compatibility we must disable `body_size` completely.
Promise chaining the emscripten module `then` function breaks it badly,
causing an infinite loop.
I'm unsure about the source of the issue, but most likely at this point
is due to the old emscripten version (I remember very old html5 builds
having issue with promise chaining too).
With this commit, we no longer use the module as a promise, and
instantiate it using `Promise` objects directly for compatibility.
`WebAssembly.instantiateStreaming` requires the mime-type to be
`application/wasm`, but some servers (including most debug servers) do
not provide the content-type header.
This commit forces it via JavaScript, by creating a `Response` object
with the `wasm` content, and explicitly defined `content-type` header.
This is beacuse on Android these values are already in m/s^2 while on
iOS and UWP they are in g. This just makes the behaviour consistent on
all platforms.
Exposes AddressSanitizer support in MSVC compiler. Can be installed via individual
components in the Visual Studio 2019 Installer.
Disabled by default. Compile the engine with `scons use_asan=yes`.
With a very nice hack, a new hidden configuration option that delays
dropped files removal at exit.
This still leaks while the project manager is running, but will clear
memory as soon as it exits or load something.
(reminder, dropped files are reguarly removed after the signal is
emitted specifically to avoid leaks, but I prefer hacking the HTML5
config then the project manager).
Added as an export option "Experimental Virtual Keyboard".
There is no zoom, so text/line edit must be in the top part of the
screen, or it will get hidden by the virtual keyboard.
UTF8/Latin-1 only (I think regular UTF-8 should work out of the box in
4.0 but I can't test it).
It uses an hidden textarea or input, based on the multiline variable,
and only gets activated if the device has a touchscreen.
This could cause problems on devices with both touchscreen and a real
keyboard (although input should still work in general with some minor
focus issues). I'm thinking of a system to detect the first physical
keystroke and disable it in case, but it might do more harm then good,
so it must be well thought.
It used to be updated before the first iteration, causing the
window/viewport size values to be incorrect during the initialization
phase (e.g. during the first `_ready` notification).
This allows to install it as an app, and provide offline support (after
the first run).
Practically, this boils down to adding a JSON file as a manifest, an
offline page to be displayed when the cached files are not avaialble,
and a JS file to cache resources and return them.
The reason for the "first run requirements" is that some browsers, will
emit an "install" by just visiting the page (to see if the JS code is
compatibile), and we do not want to force casual visitors to just
download the 10 MiB+ compressed editor WebAssembly file without pressing
the start button.
Special thanks to Hugo Locurcio (Calinou) for the initial work.
To avoid trying to do PRIME detection on fake `libGL.so` as used by e.g.
Renderdoc or Primus, we skip detection if there's a `libGL.so` in
`LD_LIBRARY_PATH`... and our luck is that Steam defines it and includes
system paths too, thus the actual system `libGL`... 🤦
So if we detect Steam, we skip this check.
Co-authored-by: Hein-Pieter van Braam-Stewart <hp@tmm.cx>
This allow the loading bar to be much more reliable, even in cases where
realible stream loading status is not detectable (server-side
compression, chunked encoding).
The problem happened on methods `screen_get_position`,
`screen_get_usable_rect` and `window_set_current_screen` when they were
passed a negative screen value.
Fixes:
- #46184
- #46185
- #46186
We used to have it like `$GODOT_VERSION` which caused inconsistencies
between different scons versions when substituting it.
It's now `@GODOT_VERSION@`, which is safe on both scons3 and scons4.
A template for `jsdoc` that generat the HTML5 public classref.
The script can be run via `npm run docs` to print to stdout.
You can dry run via `npm run docs -- --d dry-run` or write to file via
`npm run docs -- -d /path/to/file.rst`
Also update Makefile in `doc/` and add dry run test to CI.
When using use_static_cpp we want to statically link with atomic as well
to make sure we don't incur any new runtime dependencies.
Scons doesn't quite support this so we do this little trick.
According to the LLVM documentation when using GNU's libstdc++ clang
will not automatically link with -latomic. This is necessary since we
merged c++11 atomics support.
This fixes linking using Clang on Linux
This #define's older inttypes to their newer versions and #includes
<stdint.h> in the generated files. This will help with older
glibc/compiler versions using headers generated on newer systems.
This closes#46223
Three canvas resize policies:
- `None`: Godot window settings are ignored.
- `Project`: Godot handles the canvas like a native app (resizing it
when setting the window size).
- `Adaptive`: Canvas size will always adapt to browser window size.
Use `None` if you want to control the canvas size with custom JavaScript
code.
- Based on C++11's `atomic`
- Reworked `SafeRefCount` (based on the rewrite by @hpvb)
- Replaced free atomic functions by the new `SafeNumeric<T>`
- Replaced wrong cases of `volatile bool` by the new `SafeFlag`
- Platform-specific implementations no longer needed
Co-authored-by: Hein-Pieter van Braam-Stewart <hp@tmm.cx>
-Advanced Settings toggle also hides advanced properties when disabled
-Simplified Advanced Bar (errors were just plain redundant)
-Reorganized rendering quality settings.
-Reorganized miscelaneous settings for clean up.
In addition, add support for scaling and applying filter to the splash screen on Android.
One limitation of the api being used is that the splash screen aspect ratio is not maintained when it's scaled up.
This is what GitHub Actions now provide and they removed the previous 21.3.6528147.
A bit annoying to have our hand forced this way but it's still 21.x so should be good
to upgrade.
It appears that we can get a fun circle dependency on a shared object on
some system configurations causing issues with our 'fake' function
pointer names. This can lead to a crash.
The new wrapper generator renames all the symbols so this can't happen
anymore. See https://github.com/hpvb/dynload-wrapper/commit/704135e
This closes#46140
This makes it possibly to run Linux binaries compiled with udev support on
Linux systems which do not provide udev (typically systemd-less distros).
If udev is missing, we fall back to parsing `/dev/input` like when compiled
without udev support (`udev=no`).
Also adding some verbose debug statements to know which method we're using
when debugging Linux joypad issues.
The libudev so wrappers were generated on Mageia 8 with libudev 246.9 using
https://github.com/hpvb/dynload-wrapper:
```
./generate-wrapper.py --include /usr/include/libudev.h --sys-include '<libudev.h>' \
--soname libudev.so.1 --init-name libudev --omit-prefix gnu_ \
--output-header libudev-so_wrap.h --output-implementation libudev-so_wrap.c
```
By generating stubs using https://github.com/hpvb/dynload-wrapper we
can dynamically load libpulse and libasound on systems where it is available.
Both are still a build-time requirement but no longer a run-time dependency.
For maintenance purposes the wrappers should not need to be re-generated
unless we want to bump pulse or asound to an incompatible version. It is
unlikely we will want to do this any time soon.
This closes#20978
`OS.get_screen_scale` will now return the `window.devicePixelRatio`
value, `OS.get_screen_dpi` uses CSS media queries to find approximate
DPI value for the current display.
`OS.get_screen_size` also return the actual screen size (not the CSS
pixel size).
-Rendering server now uses a split RID allocate/initialize internally, this allows generating RIDs immediately but initialization to happen later on the proper thread (as rendering APIs generally requiere to call on the right thread).
-RenderingServerWrapMT is no more, multithreading is done in RenderingServerDefault.
-Some functions like texture or mesh creation, when renderer supports it, can register and return immediately (so no waiting for server API to flush, and saving staging and command buffer memory).
-3D physics server changed to be made multithread friendly.
-Added PhysicsServer3DWrapMT to use 3D physics server from multiple threads.
-Disablet Bullet (too much effort to make multithread friendly, this needs to be fixed eventually).
Issues addressed:
a) Axis mappings were including virtual mouse axes on NVIDIA Shield TV.
The virtual mouse axes have the same axis numbers as the normal analog stick numbers. This was completely breaking joypad support on NVIDIA Shield TV.
b) Joypads were being tracked in a List with the index in the list being treated as the Godot device id.
If a device were to be removed, any device later in the list would be shifted, potentially causing future events with the shifted joypads to have incorrect IDs according to the Godot engine.
c) Unnecessary events were being sent to the Godot engine.
A check was added (per Joystick) that will prevent sending events for all axes when only a single axis value changed.
A similar check was added for "HATs".
See #45712
