- Removed platform-specific implementations.
- Now all semaphores are in-object, unless they need to be conditionally created.
- Similarly to `Mutex`, provided a dummy implementation for when `NO_THREADS` is defined.
- Similarly to `Mutex`, methods are made `const` for easy use in such contexts.
- Language bindings updated: `wait()` and `post()` are now `void`.
- Language bindings updated: `try_wait()` added.
Bonus:
- Rewritten the `#ifdef` in `mutex.h` to meet the code style.
Namely, move the drive dropdown to just the left of the path text box and don't include the former
in the latter.
This improves the UX on Windows.
In the UNIX case, since its concept of drives is (ab)used to provide shortcuts to useful paths, its
dropdown is kept at the original location.
Main:
- It's now implemented thanks to `<mutex>`. No more platform-specific implementations.
- `BinaryMutex` (non-recursive) is added, as an alternative for special cases.
- Doesn't need allocation/deallocation anymore. It can live in the stack and be part of other classes.
- Because of that, it's methods are now `const` and the inner mutex is `mutable` so it can be easily used in `const` contexts.
- A no-op implementation is provided if `NO_THREADS` is defined. No more need to add `#ifdef NO_THREADS` just for this.
- `MutexLock` now takes a reference. At this point the cases of null `Mutex`es are rare. If you ever need that, just don't use `MutexLock`.
- Thread-safe utilities are therefore simpler now.
Misc.:
- `ScopedMutexLock` is dropped and replaced by `MutexLock`, because they were pretty much the same.
- Every case of lock, do-something, unlock is replaced by `MutexLock` (complex cases where it's not straightfoward are kept as as explicit lock and unlock).
- `ShaderRD` contained an `std::mutex`, which has been replaced by `Mutex`.
Without this patch, the following exception is thrown when the touch
screen is used: TypeError: e.getBoundingClientRect is not a function.
No touch events arrive in the engine.
From my testing, this PR fixes the issue and behaves as expected.
Tested with godot-demo-projects/misc/multitouch_view/, emscripten 1.39.8
and Firefox mobile emulator as well as FF on Android
- Renames PackedIntArray to PackedInt32Array.
- Renames PackedFloatArray to PackedFloat32Array.
- Adds PackedInt64Array and PackedFloat64Array.
- Renames Variant::REAL to Variant::FLOAT for consistency.
Packed arrays are for storing large amount of data and creating stuff like
meshes, buffers. textures, etc. Forcing them to be 64 is a huge waste of
memory. That said, many users requested the ability to have 64 bits packed
arrays for their games, so this is just an optional added type.
For Variant, the float datatype is always 64 bits, and exposed as `float`.
We still have `real_t` which is the datatype that can change from 32 to 64
bits depending on a compile flag (not entirely working right now, but that's
the idea). It affects math related datatypes and code only.
Neither Variant nor PackedArray make use of real_t, which is only intended
for math precision, so the term is removed from there to keep only float.
This attribute is now part of the standard we target so we no longer
need compiler-specific hacks.
Also enables -Wimplicit-fallthrough for Clang now that we can properly
support it. It's already on by default for GCC's -Wextra.
Fixes new warnings raised by Clang's -Wimplicit-fallthrough.
It's now available and allows us to have a better default environment,
with GCC 7.4.0 and Clang 7.
We now need GCC 7+ for C++17 support so it's more efficient to upgrade
the image than to install it on Ubuntu 16.04 (xenial).
Also fixes a couple -Wdeprecated-declarations warnings on macOS now
that we build against macOS 10.12.
As per #36436, we now need C++17's guaranteed copy elision feature to
solve ambiguities in Variant.
Core developers discussed the idea to move from C++14 to C++17 as our
minimum required C++ standard, and all agreed. Note that this doesn't
mean that Godot is going to be written in "modern C++", but we'll use
modern features where they make sense to simplify our "C with classes"
codebase. Apart from new code written recently, most of the codebase
still has to be ported to use newer features where relevant.
Proper support for C++17 means that we need recent compiler versions:
- GCC 7+
- Clang 6+
- VS 2017 15.7+
Additionally, C++17's `std::shared_mutex` (conditionally used by
`vk_mem_alloc.h` when C++17 support is enabled) is only available in
macOS 10.12+, so we increase our minimum supported version.
It was initially implemented in #5871 for Godot 3.0, but never really
completed or thoroughly tested for most platforms. It then stayed in
limbo and nobody seems really keen to finish it, so it's better to
remove it in 4.0, and re-add eventually (possibly with a different API)
if there's demand and an implementation confirmed working on all
platforms.
Closes#8770.
Due to the port to Vulkan and complete redesign of the rendering backend,
the `drivers/gles3` code is no longer usable in this state and is not
planned to be ported to the new architecture.
The GLES2 backend is kept (while still disabled and non-working) as it
will eventually be ported to serve as the low-end renderer for Godot 4.0.
Some GLES3 features might be selectively ported to the updated GLES2
backend if there's a need for them, and extensions we can use for that.
So long, OpenGL driver bugs!
- Renamed option to `builtin_vulkan`, since that's the name of the
library and if we were to add new components, we'd likely use that
same option.
- Merge `vulkan_loader/SCsub` in `vulkan/SCsub`.
- Accordingly, don't use built-in Vulkan headers when not building
against the built-in loader library.
- Drop Vulkan registry which we don't appear to need currently.
- Style and permission fixes.
-Texture renamed to Texture2D
-TextureLayered as base now inherits 2Darray, cubemap and cubemap array
-Removed all references to flags in textures (they will go in the shader)
-Texture3D gone for now (will come back later done properly)
-Create base rasterizer for RenderDevice, RasterizerRD
-Added VulkanContext
-Added an X11 implementation
-Added a rendering device abstraction
-added a Vulkan rendering device abstraction
-Engine does not work, only shows Godot logo (run it from bin/)
Calling `step()` on EditorProgress too often will slow down the
rest of the editor, so it's best avoided. This is also more consistent
with other exporters, as most of them don't report per-file progress
either.
Exporting a 2D project with ~1,100 files to Android now takes
about 10 seconds from a debug editor build instead of 65 seconds.
This closes#30850.
Changed the condition to add a length filter to make it consistent with the documentation (0 means no character limit). Otherwise the default value in LineEdit causes the virtual keyboard to be non-fonctional on Android.
Reverts the following commits:
- c81ec6f26d:
"Exposes capture methods to AudioServer, variable renames for
consistency, added documentation."
- 47c558b98a:
"Expose audio callbacks as signals."
- dabaa11b3c:
"Fix to make sure the capture buffers are deallocated at shutdown.
Silences warnings."
Some documentation improvements were kept for pre-existing methods.
See rationale for reverting these changes in #30468.
- `EditorNavigationMeshGenerator` was being registered as part of the Core API,
even after d3f48f88bb. We must make sure to
set Editor as the current ClassDB API type before creating an instance.
- The `VisualScriptEngineSingleton.constant` property has a property hint string
that's different between tools and non-tools builds. This commit makes the
hint string to no longer be set in `_bind_methods`, and to instead set it in
`_validate_property`. This way it's ignored when calculating the API hash.
- `JavaClassWrapper` is now registered in ClassDB on all platforms,
using a dummy implementation on platforms other than Android.
This fixes API portability between Android and other platforms.
- Updated `--class-db-json` command to ignore non-virtual methods that start
with an underscore (see: 4be87c6016).
Fixes compatibility with emscripten 1.39.5+ .
Most input callbacks now require a target and no longer support NULL
defaults.
This commit changes all required null targets to the expected default in
the binding phase.
Since for canvas-related callbacks there is no default, the "#canvas"
selector is used instead.
Additionally, since canvasX and canvasY event properties are no longer
supported, event positions are computed from "clientX" and "clientY" and
the "#canvas" bounding client rect.
It was removed as noted in the changelog:
https://github.com/emscripten-core/emscripten/blob/1.39.5/ChangeLog.md#v1395-12202019
> Removed `timestamp` field from mouse, wheel, devicemotion and
> deviceorientation events. The presence of a `timestamp` on these
> events was slightly arbitrary, and populating this field caused
> a small profileable overhead that all users might not care about.
> It is easy to get a timestamp of an event by calling
> `emscripten_get_now()` or `emscripten_performance_now()` inside
> the event handler function of any event.
Fixes#34648.
Add check to prevent compositor sync if the graphics driver is forcing vsync.
Fixes#35038.
(Addendum: this PR does not negatively impact users unaffected by #35038.)
- Go up was not working, simplify was used one time too much
- Added GestureHandler
- Added doubleTap to recognize open dir
- Fixed scroll where sometimes the scroll jumped between start and end when pointer was outside or on the edge of the scroll area
I'm barely scratching the surface of the changes needed to make the
--export command line interface easy to use, but this should already
improve things somewhat.
- Streamline `can_export()` templates check in all platforms, checking
first for the presence of official templates, then of any defined
custom template, and reporting on the absence of any.
Shouldn't change the actual return value much which is still true if
either release or debug is usable - we might want to change that
eventually and better validate against the requested target.
- Fix discrepancy between platforms using `custom_package/debug` and
`custom_template/debug` (resp. `release`).
All now use `custom_template`, which will break compatibility for
`export_presets.cfg` with earlier projects (but is easy to fix).
- Use `can_export()` when attempting a command line export and report
the same errors that would be shown in the editor.
- Improve error reporting after a failed export attempt, handling
missing template and invalid path more gracefully.
- Cleanup of unused stuff in EditorNode around the export workflow.
- Improve --export documentation in --help a bit.
Fixes#16949 (at least many of the misunderstandings listed there).
Fixes#18470.
Unify pack file version and magic to avoid hardcoded literals.
`version.py` now always includes `patch` even for the first release in
a new stable branch (e.g. 3.2). The public name stays without the patch
number, but `Engine.get_version_info()` already included `patch == 0`,
and we can remove some extra handling of undefined `VERSION_PATCH` this
way.
Co-authored-by: Rémi Verschelde <rverschelde@gmail.com>
Happy new year to the wonderful Godot community!
We're starting a new decade with a well-established, non-profit, free
and open source game engine, and tons of further improvements in the
pipeline from hundreds of contributors.
Godot will keep getting better, and we're looking forward to all the
games that the community will keep developing and releasing with it.
Provides access to a MulticastLock.
As specified by the Android API, broadcast/multicast packets may be
filtered on some phones unless the application explicitly acquires
a "MulticastLock".
On iOS devices without a physical home button iOS
shows a home indicator instead. This is often in the
way of the UI or the game.
Added a project setting to disable hidden home indicator.
The default value is to hide the home indicator
The new 'split_libmodules=yes' option is useful to work around linker
command line size limitations when linking a huge number of objects.
We're currently over 64k chars when linking libmodules.a on Windows
with MinGW, which triggers issues as seen in #30892.
Even on Linux, we can also reach linker command line size limitations
by adding more custom modules.
We force this option to True for MinGW on Windows, which fixes#30892.
Additional changes to lib splitting:
- Fix linking of the split module libs with interdependent symbols,
hacking our way into LINKCOM and SHLINKCOM to set the `--start-group`
and `--end-group` flags.
- Fix Python 3 compatibility in `methods.split_lib()`.
- Drop seemingly obsolete condition for 'msys' on 'posix'.
- Drop the unnecessary 'split_drivers' as the drivers lib is no longer
too big since we moved all thirdparty builds to modules.
Co-authored-by: Hein-Pieter van Braam-Stewart <hp@tmm.cx>
We no longer compile for armv7 and x86 in the official export templates,
as those architectures are no longer relevant for iOS.
If users really want to support armv7 (used on devices from before
September 2013, e.g. iPhone 5), they can still build their own templates
and toggle the option.
We might remove the option altogether in a later release to avoid the
confusion for users that might tick the checkbox without having compiled
their own templates.
Fixes#34135.
The option is needed when using the 'fastcomp' backend (default before
1.39.0), and must not be defined when using 'upstream' (new default).
So we define it conditionally to support both backends.
Follow-up to #30751.
This makes it possible to know whether the window is focused
at a given time, without having to track the focus state manually
using `NOTIFICATION_WM_FOCUS_IN` and `NOTIFICATION_WM_FOCUS_OUT`.
This partially addresses #33928.
This allows setting the `read_chunk_size` of the internal HTTPClient.
This is important to reduce the allocation overhead and number of file
writes when downloading large files, allowing for better download speed.
Third-party platforms (e.g. console ports) need to be able to
disable JIT support in the regex module too, so it can't be
hardcoded in the module SCsub. This is cleaner this way anyway.
Fixes#19316.
Upstream Emscripten changed this in 1.39.1+, so IDBFS is no longer
included by default and has to be linked manually.
The explicit linking doesn't seem to be problematic on earlier
versions (tested `1.38.47-upstream`).
Fixes#33724.
A change in upstream Emscripten 1.39.1+ made our buildsystem error
out where it was previously only issuing a warning:
```
[ 5%] Linking Static Library ==> main/libmain.javascript.opt.bc
shared:WARNING: Assuming object file output in the absence of `-c`, based on output filename. Please add with `-c` or `-r` to avoid this warning
Ranlib Library ==> main/libmain.javascript.opt.bc
/opt/emsdk/upstream/bin/llvm-ranlib: error: unable to load 'main/libmain.javascript.opt.bc': file too small to be an archive
```
As advised on emscripten-core/emscripten#9806, we should be using
`emar` here to create the static library and not `emcc`.
This was apparently done to workaround Emscripten issues in the past,
but evidently this is no longer necessary.
The rest of the `env` redefinitions should probably be re-assessed
against the current state of Emscripten.
Fixes#33374.
Since most browsers no longer allow making async requests from a page
loaded from `file://`, we now need a proper HTTP server to load the
exported HTML5 game.
This should also allow us to get the debugger to work over a WebSocket
connection.
The description appears when hovering over the one-click-deploy button (top-right). This information helps the user distinguish between their devices if multiple are connected or if the same device is connected by both usb and tcpip (two entries in the list for the same device).
Avoid using adb reverse if deploying with adb tcpip.
This still can fail if the user is attempting to debug over usb and has connected their device over BOTH usb and tcpip. I'm not sure how we would detect that problem in advance though.
- Add or remove the necessary subdirectorires to the includes to remove
dependency on the editor directory being in the build's include path.
- Ensure includes in modified files conform to style guideline.
- Remove editor from the build include path.
A better fix would be to make Godot's export code properly parse the
tag over multiple lines (and maybe even use XMLParser instead of doing
it ad-hoc?).
As for the APK names, we could alternatively pick the first .apk found
in the `debug` and `release` folders without expecting a specific name.
Fixes#32414.
Example: To generate for the `release` build target and for the `armv7`, `arm64v8` and `x86` architectures, run the commands:
```
cd godot
scons -j4 platform=android target=release android_arch=armv7
scons -j4 platform=android target=release android_arch=arm64v8
scons -j4 platform=android target=release android_arch=x86
cd platform/android/java
./gradlew generateGodotTemplates
```
Notes:
- The generated build templates will be located in the `godot/bin` directory (i.e: `android_debug.apk`, `android_release.apk`, `android_source.zip`).
- The gradle command will only generate templates for the target(s) with available native shared libraries. For example, running the commands above will only generate the `android_release.apk` and `android_source.zip` files.
To delete the generated artifacts, the following commands can be used:
```
cd platform/android/java
./gradlew cleanGodotTemplates
```
Fixes#32168.
Previously we were returning all key up and key down messages as unhandled to the OS. This was resulting in crashes on certain keypresses (left cursor), for undetermined reason.
This PR defaults all key up and keydown messages to be returned as handled by Godot, except those explicitly coded as exceptions (currently volume keys only).
This fades out messages originating from the editor to make messages
printed by the project stand out more.
This also tweaks wording in some editor messages for consistency.
The application module `app` serves double duties of providing the prebuilt Godot binaries ('android_debug.apk', 'android_release.apk') and the Godot custom build template ('android_source.zip').
The language didn't make it clear that it's installing a *source* template
to the project folder, for later use when compiling custom APKs.
Fixes#28736.
It does check its permission every `vibrate_handheld()` calls.
Vibrate permission is added by checking it on export settings.
And there are some changes for deprecated method.
It had been synced with style changes (spaces -> tabs), not sure why
I accepted to merge it this way back then...
Synced with eb57657f66,
same as before.
Custom-changes will be reapplied in the next commit, if relevant.
So far we left most temporary files lying around, so this attempts to
fix that.
I added a helper method to DirAccess to factor out the boilerplate of
creating a DirAccess, checking if the file exists, remove it or print
an error on failure.
WM_MOUSEWHEEL and WM_MOUSEHWHEEL report mouse coordinates relative to
the screen (see lParam in [1]), rather than to the window like the rest
of the mouse events.
The current code already makes adjustments to take that into account.
However, it only makes the adjustments if the mouse is not captured, and
the coordinates are always relative to the screen regardless of whether
the mouse is captured or not, so let's fix the code to always
consistently apply the adjustments.
This fixes#29559.
[1] - https://docs.microsoft.com/en-us/windows/win32/inputdev/wm-mousewheel
As of 3.1 and later, we have too many thirdparty C++ dependencies
and some internal uses of `new` and `delete` too for it to make
sense to build without the STL on Android.
The option has been broken since 3.0, and the "System STL" that we
relied on for basic support of `new` and `delete` is likely to be
dropped from the NDK:
https://android.googlesource.com/platform/ndk/+/ndk-release-r20/docs/BuildSystemMaintainers.md#System-STL
Also added support for SCons project-absolute paths (starting with #) and
warning about duplicates in add_source_files(), and fixed
default_controller_mappings.gen.cpp being included twice after first build
due to *.cpp globbing.
Part of #30270.
It is not supported in Emscripten's `latest-upstream` LLVM backend,
and doesn't seem necessary in the `latest` backend either.
It was initially added in #22857 to solve a compilation error with the latter.
Part of #30270.
Emscripten is apparently changing the variables in its config file,
causing potential breakage of our build system.
Binaries of the latest/latest-upstream releases are located in a
subfolder of BINARYEN_ROOT called emscripten.
Binaries of the other releases (e.g. sdk-1.38.31-64bit) are instead
placed under the EMSCRIPTEN_ROOT folder.
This PR checks if BINARYEN_ROOT has a subfolder called emscripten, if
that does not exists, it falls back to checking the EMSCRIPTEN_ROOT.
This way we give precedence to the new releases, given that activating
multiple releases sequentially might result in having mismatching
BINARYEN_ROOT and EMSCRIPTEN_ROOT.
For clarity, assign-to-release idiom for PoolVector::Read/Write
replaced with a function call.
Existing uses replaced (or removed if already handled by scope)
Currently, the console appears when running OS.execute in an exported project,
but not in the editor. This change prevents it from appearing in either.
Only affects console applications.
The basic point is as in 2.1 (appending the PCK into the executable), but this implementation also patches a dedicated section in the ELF/PE executable so that it matches the appended data perfectly.
The usage of integer types is simplified in existing code; namely, using plain `int` for small quantities.
It's the recommended way to set those, and is more portable
(automatically prepends -D for GCC/Clang and /D for MSVC).
We still use CPPFLAGS for some pre-processor flags which are not
defines.
Godot core needs MD5/SHA256/AES/Base64 which used to be provided by
separate libraries.
Since we bundle mbedtls in most cases, and we can easily only include
the needed sources if we so desire, let's use it.
To simplify library changes in the future, and better isolate header
dependencies all functions have been wrapped around inside a class in
`core/math/crypto_base.h`.
If the mbedtls module is disabled, we only bundle the needed source
files independently of the `builtin_mbedtls` option.
If the module is enabled, the `builtin_mbedtls` option works as usual.
Also remove some unused headers from StreamPeerMbedTLS which were
causing build issues.
This PR introduces support for ARKit to the iOS version of Godot.
ARKit is Apples Augmented Reality platform.
This PR brings in support for ARKit 1.0 and implements a few ARKit 2.0 features.
It requires iOS 11 to run but should not prevent Godot from running on older versions as long as ARKit remains unused.
The rationale for keeping those shared by default is that they're typical
dependencies found on any Linux system, and it saves compilation time and
binary size to link their dynamically.
But since official builds default to all-builtin, and Debian/Ubuntu still
don't have libpng16 (which we now require) readily available on all their
supported releases, it's simpler to bundle all the things.
This does not change the fact that those dependencies *can* be unbundled
on Linux, it's only the default option changing.
Fixes#17004
Currently the keydown and keyup messages are handled with method like this:
if ((source & InputDevice.SOURCE_JOYSTICK) == InputDevice.SOURCE_JOYSTICK
|| (source & InputDevice.SOURCE_DPAD) == InputDevice.SOURCE_DPAD
|| (source & InputDevice.SOURCE_GAMEPAD) == InputDevice.SOURCE_GAMEPAD) {
// joystick input
}
else
{
// keyboard input
}
The constant for SOURCE_DPAD is 513
10 0000 0001
and the constant for SOURCE_KEYBOARD is 257
1 0000 0001
However, rather confusingly, for many keyboards the source sent by android is 769
11 0000 0001
Thus the keyboard is passing the check as being a DPAD and being processed as a joystick rather than keyboard. This PR handles the specific case of 769, allowing input from physical keyboards.
This is an editor setting and its value can also be toggled
using an entry in the Editor toolbar. The console will still
appear briefly when starting the project manager or editor,
as it's still compiled as console application.
Does not impact exported games, which will still run without
console in release and with console in debug mode.
A project setting or export option could be added to disable
it in debug mode if there's demand for it, but that would
greatly reduce the usefulness of debug builds if Windows users
can no longer report error and crash messages.
Fixes#17889.
Co-authored-by: Rémi Verschelde <rverschelde@gmail.com>
Recent Emscripten SDK versions seem to only include the
`BINARYEN_ROOT` variable in the Emscripten configuration file,
whereas the platform's `detect.py` only looked at `EMSCRIPTEN_ROOT`.
Wrapped libpng usage in a pair of functions under PNGDriverCommon,
which convert between Godot Image and png data.
Switched to libpng 1.6 simplified API for ease of maintenance.
Implemented ImageLoaderPNG and ResourceSaverPNG in terms of
PNGDriverCommon functions.
Travis, switched to builtin libpng (thus builtin freetype and zlib also)
so we can build on Xenial.
Warnings raised by Emscripten 1.38.0 and MinGW64 5.0.4 / GCC 8.3.0.
JS can now build with `werror=yes warnings=extra`.
MinGW64 still has a few warnings to resolve with `warnings=extra`,
and only one with `warnings=all`.
Part of #29033 and #29801.
This is a new singleton where camera sources such as webcams or cameras on a mobile phone can register themselves with the Server.
Other parts of Godot can interact with this to obtain images from the camera as textures.
This work includes additions to the Visual Server to use this functionality to present the camera image in the background. This is specifically targetted at AR applications.
It's not necessary, but the vast majority of calls of error macros
do have an ending semicolon, so it's best to be consistent.
Most WARN_DEPRECATED calls did *not* have a semicolon, but there's
no reason for them to be treated differently.
This updates our local copy to commit 5ec8339b6fc491e3f09a34a4516e82787f053fcc.
We need a recent master commit for some new features that we use in Godot
(see #25543 and #28909).
To avoid warnings generated by Bullet headers included in our own module,
we include those headers with -isystem on GCC and Clang.
Fixes#29503.
Being async, the first time a value is pasted GUI elements will still
return the previous one.
This at least until 'clipboardchange' window event gets implemented by
user agents.
Listen to paste events to update local clipboard.
CTRL+V still not working out of the box.
To do that, We would need to change how we handle keypress, most likely
making it worse and less safe. In the end, I'm not sure we can fix it
properly for now. Maybe in the future, with the Clipboard API, support
of which is still pretty limited on chrome, and only available to
extensions in Firefox.
For now, you can paste via:
- Browser bar -> Edit -> Paste.
- Middle mouse click (Linux only, copies secondary clipboard).
And THEN press CTRL+V
text=auto works well in Git 2.10+ but it's broken in previous versions,
which are still used in production on e.g. Ubuntu 16.04 LTS.
Also fix a couple missed text files with CRLF terminators.
.bat files likely require it to be processed properly on Windows,
but core.autocrlf should take care of converting them on the fly
when checking out on Windows.
In x11, windows and osx crash handlers, check project settings exists
before looking up the crash handler message setting.
Avoids crashing the crash handler when handling a crash outside project
settings lifetime. Instead omitting the configurable message and
continuing with trace dump.
Those were disable to keep size small, and on Android avoid the dependency on the STL,
but for tools build (editor) this is not really a concern.
Note: as of today it's not possible to build tools=yes for those platforms, but this
change is one of the necessary steps to enable it.
Fixes#25262.
Reasoning: ID is not an acronym, it is simply short for identification, so it logically should not be capitalized. But even if it was an acronym, other acronyms in Godot are not capitalized, like p_rid, p_ip, and p_json.
Include paths are processed from left to right, so we use Prepend to
ensure that paths to bundled thirdparty files will have precedence over
system paths (e.g. `/usr/include` should have lowest priority).
This is the same as #23542 (Fix binaries incorrectly detected as shared
libraries on some linux distros) but for Clang. It should be fine with
Clang 4 or higher.
This adds ThinLTO support when using Clang and the LLD Linker, it's
turned off by
default.
For now only support for Linux added as ThinLTO support on other
platforms may still be buggy.
Many contributors (me included) did not fully understand what CCFLAGS,
CXXFLAGS and CPPFLAGS refer to exactly, and were thus not using them
in the way they are intended to be.
As per the SCons manual: https://www.scons.org/doc/HTML/scons-user/apa.html
- CCFLAGS: General options that are passed to the C and C++ compilers.
- CFLAGS: General options that are passed to the C compiler (C only;
not C++).
- CXXFLAGS: General options that are passed to the C++ compiler. By
default, this includes the value of $CCFLAGS, so that setting
$CCFLAGS affects both C and C++ compilation.
- CPPFLAGS: User-specified C preprocessor options. These will be
included in any command that uses the C preprocessor, including not
just compilation of C and C++ source files [...], but also [...]
Fortran [...] and [...] assembly language source file[s].
TL;DR: Compiler options go to CCFLAGS, unless they must be restricted
to either C (CFLAGS) or C++ (CXXFLAGS). Preprocessor defines go to
CPPFLAGS.
On Windows, when "Language for non-Unicode programs" were set to "Japanese (Japan)", MSVC would by default use Shift JIS (code page 932) to interpret source files, which would result in test_string failing to compile because of characters in `test_34()`. Forcing utf-8 for MSVC fixes the issue
It seems to stay compatible with formatting done by clang-format 6.0 and 7.0,
so contributors can keep using those versions for now (they will not undo those
changes).
Added constructor and assignment operator for CharString
from const char* to simplify memory management when working with
utf8/ascii strings for APIs taking char*.
Reworked OS_X11::set_context to use CharString and avoid some manual
memory management.
When getting system directories for Windows, we currently use
SHGetFolderPathW. This is a deprecated function and doesn't support
"Downloads" folders.
As a replacement, this commit uses the newer SHGetKnownFolderPath
function, which is supported since Windows Vista. Godot 3.0 only
supports Windows 7+, so we don't need to use SHGetFolderPathW for
backwards compatibility.
Fixes#26876
GLES2 is not designed to be a drop-in replacement for the GLES3 backend,
so the fallback mode has to be used knowingly. It *can* make sense for
simple projects which make sure to handle the differences between both
rendering backends, but most users should stick to one supported backend.
By making it opt-in, we can now use this parameter to define whether to
export ETC textures to Android and iOS when using GLES3 + Fallback.
When using GLES3 without Fallback on Android, set the proper min GLES
version in the AndroidManifest.
Also made the option boolean and renamed it for clarity and to avoid
conflict with the previous String option (which would always evaluate as
"true" otherwise).
Fixes#26569.
For HTML5, we need to support S3TC if running on desktop,
and ETC or ETC2 for mobile, so make this explicit.
Add logic to check for ETC2 support on GLES3,
and remove incorrect ETC feature for GLES3 on Android.
Fix ETC check invalidating templates on HTML5.
Fixes#26476.
We've been defaulting to WASAPI since 3.0 and it's superior to RtAudio
in all aspects.
Obsoletes and closes#25503.
Also enable WINMIDI on MinGW, this had been missed initially.
Fix os_windows.cpp and crash_handler_windows.cpp which had weird
dependencies on RtAudio.h's includes (ugh).
It seems that bumblebee doesn't like us creating multiple GL contexts
to avoid this we now detect whether we're running with this software
and don't do anything.
Also drop some unused files.
Renamed:
- `platform/iphone/sem_iphone.h` -> `semaphore_iphone.h`
(same for `osx`)
- `platform/uwp/gl_context_egl.h` -> `context_egl_uwp.h`
- in `platform/windows`: `context_gl_win.h`, `crash_handler_win.h`,
`godot_win.cpp`, `joypad.h` and `key_mapping_win.h` all renamed to
use `windows`. Some classes renamed accordingly too.
- `EditorExportAndroid` and `EditorExportUWP` renamed to
`EditorExportPlatformAndroid` and `EditorExportPlatformUWP`
- `power_android` and `power_osx` renamed to `PowerAndroid` and
`PowerOSX`
- `OSUWP` renamed to `OS_UWP`
Dropped:
- `platform/windows/ctxgl_procaddr.h`
This allows most demos to run without any ubsan or asan errors. There
are still some things in thirdpart/ and some things in AudioServer that
needs a look but this fixes a lot of issues. This should help debug less
obvious issues, hopefully.
This fixes#25217 and fixes#25218
We fork off twice once with and once without DIR_PRIME=1 set. We
then use the vendor string to determine what GPU to use.
We prefer (in order)
1) AMDGPU/AMDGPU-PRO/NVidia non-free driver
2) Intel driver
3) Nouveau
4) Software rendering
If a driver can't be detected it will default to DRI_PRIME=0
While looking into a different issue, I've noticed that Visual Studio Intellisense does not work well for Godot project when using Windows Vista+ APIs (e.g. CreateThreadpool), i.e. it does not recognise the APIs because they are defined in Windows header files for Vista+ only.
This is because the WINVER and _WIN32_WINNT symbols don't have their values set in the generated Godot project file. This fixes the problem by setting the values when generating the project file.
Make the sanitizer names more explicit (use_ubsan, use_asan, use_lsan).
Comment has been adjusted to include GCC as supported compiler for these
and exclude -fno-omit-frame-pointer option (should not cause any
problems).
From August 1, 2019, Google Play requires that all new apps and app updates
include 64-bit versions, so we enable ARM64 by default.
IINM support for x86 and x86_64 is still be optional, so not enabling them
out of the box.
Part of #25030.
Godot supports many different compilers and for production releases we
have to support 3 currently: GCC8, Clang6, and MSVC2017. These compilers
all do slightly different things with -ffast-math and it is causing
issues now. See #24841, #24540, #10758, #10070. And probably other
complaints about physics differences between release and release_debug
builds.
I've done some performance comparisons on Linux x86_64. All tests are
ran 20 times.
Bunnymark: (higher is better)
(bunnies) min max stdev average
fast-math 7332 7597 71 7432
this pr 7379 7779 108 7621 (102%)
FPBench (gdscript port http://fpbench.org/) (lower is better)
(ms)
fast-math 15441 16127 192 15764
this pr 15671 16855 326 16001 (99%)
Float_add (adding floats in a tight loop) (lower is better)
(sec)
fast-math 5.49 5.78 0.07 5.65
this pr 5.65 5.90 0.06 5.76 (98%)
Float_div (dividing floats in a tight loop) (lower is better)
(sec)
fast-math 11.70 12.36 0.18 11.99
this pr 11.92 12.32 0.12 12.12 (99%)
Float_mul (multiplying floats in a tight loop) (lower is better)
(sec)
fast-math 11.72 12.17 0.12 11.93
this pr 12.01 12.62 0.17 12.26 (97%)
I have also looked at FPS numbers for tps-demo, 3d platformer, 2d
platformer, and sponza and could not find any measurable difference.
I believe that given the issues and oft-reported (physics) glitches on
release builds I believe that the couple of percent of tight-loop
floating point performance regression is well worth it.
This fixes#24540 and fixes#24841
- The `cpu-features.{c,h}` code was only used by chance by the webm
(libvpx) code, so I moved it there. It was actually introduced before
that and wasn't in use, and libvpx just happened to be able to
compile thanks to it being bundled.
It could potentially be compiled on the fly from the Android NDK, but
since we plan to replace the webm module by a GDNative plugin in the
near future, I went the bundling route.
- `ifaddrs_android.h` is already provided in the Android NDK as
`ifaddrs.h`, same as on other Unixes. Yet we cannot use it until we
up the min API level to 24, where `getifaddrs` is first defined.
I moved the files to `thirdparty/misc` and synced them with upstream
WebRTC (only indentation changes and removal of `static` qualifiers).
Also removes dropped thirdparty files from COPYRIGHT.txt after changes
in #24105 and #24145.
By introducing an intermediate proxy class for the array subscript
operator for String and CharString we can control better when CowData
will actually CoW.
This should improve performance of String usage for most cases.
The current system for capturing the mouse and generating motion events on X11
has issues with inaccurate and lopsided input. This is because both
XQueryPointer and XWarpPointer work in terms of integer coordinates when the
underlying X11 input driver may be tracking the mouse using subpixel
coordinates. When warping the pointer, the fractional part of the pointer
position is discarded.
To work around this issue, the fix uses raw motion events from XInput 2. These
events report relative motion and are not affected by pointer warping.
Additionally, this means Godot is able to detect motion at a higher resolution
under X11. Because this is raw mouse input, it is not affected by the user's
pointer speed and acceleration settings. This is the same system as SDL2 uses
for its relative motion.
Multitouch input on X requires XInput 2.2. Raw motion events require
XInput 2.0. Since 2.0 is old enough, this is now the minimum requirement to
use Godot on X.
This mirrors behavior on Windows and MacOSX where Godot tries to default
to a dGPU if it exists. This doesn't work for Nvidia optimus yet but
this can maybe be added later.
Contrarily to what #23434 assumed, this is not a memory leak,
the VisualServerRaster instance is passed as a parameter to
VisualServerWrapMT's constructor.
Fixes#23437.
