As discussed in issues #1479 and #9782, choosing the up axis (which is Y in Godot) as the axis of the last (or first) rotation is helpful in practical use cases.
This also aligns Godot's convention with Unity, helping with a smoother transition for people who are used to working with Unity (issue #9905).
Internally, both XYZ and YXZ functions are kept, for potential future applications.
Plus:
- An allocation is counted only after checking its success.
- Max usage is updated after growing reallocs as well.
- Drop unused header.
- Changed the 0xFFF.. at get_mem_available() to -1 with a comment telling it's the same, but more universal.
There was no constructor for Ref from const pointer, so compiler decided
to construct Variant from pointer and then construct Ref from Variant
which turned it into NULL, because the Variant had null ref field.
When looking up a symbol from a library, previously an error was
shown when the symbol did not exist. That caused confusion when the
lookup was completely optional.
This adds a new parameter to that method so that those errors can
be handled manually if needed.
This also changes Reference::unreference() to always invoke
refcount_decremented. Previously it was not invoked until the count
reached zero due to short-circuit evalution of boolean expressions.
Notice: GDScript tokenizer used the old PoolFloatArray name.
Renamed PoolFloatArray to PoolRealArray.
Moved "project_settings.h" down one line to comply with the clang-format rules.
Fixes#9638
Closed pull request #9714 because I messed up with commits, sorry!
I'm working on Nim bindings and Nim GC needs to know the stack
boundaries to check whether certain pointers are located on the stack
or in the heap. This commit adds godot_get_stack_bottom procedure
to gdnative module which returns pointer to the stack bottom of the
main thread. Later on this may be improved to return stack bottom of
the current thread.
- Remove use of non-builtin overloaded `InterlockedExhangeSubtract()` for Windows. Moreover, it was wrongly being used as if it hadn't a 64-bit version instead of being overloaded.
- Implement exchange-if-greater (as a CAS loop).
The script system does not provide information about specific int
sizes, so we should establish convention to use the largest size
(64 bits). For real types double is always used.
-Added new 3D stream player node
-Added ability for Area to capture sound from streams
-Added small features in physics to be able to properly guess distance to areas for sound
-Fixed 3D CollisionObject so shapes are added the same as in 2D, directly from children
-Fixed KinematicBody API to make it the same as 2D.
The public API looks the same but with added overloads.
The implementation uses templates or overloads depending on the context
to avoid code duplication.
Using a magic value for memory allocated but uninitialized and another one for memory about-to-be-released. Helps in debugging unitialized members, dangling pointerts, etc.
Disabled by default. Can be enabled for debug builds by defining `DEBUG_MEMORY_TAGGING`.
The allocation count is managed atomically and where it actually should
change (for instance, not counting an allocation before its success has
been checked).
Bonus: Improve readability of the pre-pad checks.
Plus the addition of some convenience CRASH_* error macros.
Plus transient avoidance of the flood of warnings emitted by Clang when checking 'this' for NULL.
Plus explanation about the do-while(0) loop in some error macros.
-Added ability to set/get a field in GetSet, as well as assignment ops
-Added a Select node
-Fixed update bugs related to variable list and exported properties, closes#9458
Some methods require a const Ref<T> argument,
the ptrcall method wrappers cast `void *` to the
apropriate types. The problem is that there is no `Ref(const T *)`
constructor, but since Ref modifies the refcount of a Reference
anyway there's no point in a const version.
The problem is that with a `const T *` constructor call, the
argument gets converted to Variant first and loses all the
reference information, resulting in a null reference as the
argument to the constructor.
1. Let r1 be the source rectangle clipped against the entire source image rectangle.
2. Let r2 be r1 offset by p_dest, clipped against the entire destination image rectangle.
3. Copy pixels from r1 to r2.
Templates will be loaded from .godot/script_templates
For now they're disabled for GDNative.
Ideas for further improvements:
- Add a "Save as Template" option to the script editor, as it can normally only save to res://
- Support more placeholders / custom placeholders
zstd has much better compression speed and ratio, and better decompression speed than currently available methods.
Also set zstd as the default compression method for Compression as well as FileAccessCompressed functions.
Quite a while ago I made a commit (131631b) where I did a weird
thing to fix compilation with PTRCALL_ENABLED. And I couldn't
sleep because of this after all these months. So here is the
proper version.
-Added default environment editor setting
-Added environment created by default in new projects
-Removed default light and ambient from spatial editor, to make the editor more PBR compliant
The `InputEvent::is_action(pressed|released)` methods weren't implemented yet.
Also fixed a typo in `InputDefault` that prevented `Input.is_action(pressed|released)` from working.
-Improvements to texture importer
-Proper detection of S3TC compression modes, and added all modes to Image
-Fixes to non-power of 2 compressed textures, which should all be supported by GLES3
This is a necessary condition for finding optimal solutions.
This is achieved by simply requiring/ensuring that no weights are smaller than 1.
Fixes#8584.
Make EditorFileDialog, FileDialog and EditorFileSystemDirectory alphanumerical sorting more natural
Added a new method 'naturalnocasecmp_to' and comparator 'NaturalNoCaseComparator' to String.
Fixes#8712.
Slimmed down variant from the reverted #8375.
The rationale behind the name change is to give Godot's project file a unique
extension (".godot") that can be registered on the OS to be associated with
the Godot binary (OS registration not implemented here).
This PR also adds the possibility to start the game or editor if launched
with the project.godot passed as argument, which paves the way for allowing
a similar behaviour on a double-click in the OS file manager (code originally
by @Hinsbart).
Closes#6915.
"ALL IS GOOD" was a lie.
In particular, removes verbose "path not recognized" false positive.
The actual logic is to (somewhat naively) check all ResourceFormatLoaders
and to pick the first good match, so no need to warn about the formats
that do not match the type hint.
Fixes a bug where the VERSION_PATCH define is not yet in scope if
typedefs.h is included before version.h at compilation time.
(cherry picked from commit 3b687c5474)
PCG32 doesn't like small seeds, which leads to zero random values (prior to #7532, zero values were handled as special cases).
Use a large default seed, and also add a shift in Math::randomize.
Fixes#8423.
Notable changes:
- Now ScriptLanguages have the option to override the global external editor setting.
If `ScriptLanguage::open_in_external_editor()` returns `ERR_UNAVAILABLE` (which it does by default), then the global external editor option will be used.
- Added formatting to the external editor execution arguments. Now it's possible to write something like this: `{project} -g {file}:{line}:{col}`.
- `VisualScript::get_member_line()` now can return the line of functions (well, it returns the id of the _Function_ node of the function). I guess there is nothing else we can get a "line" from.
Fixes:
- Fixes a bug where `ScriptEditor::script_goto_method()` would not work if the script is not already open in the built-in editor.
- Fixes wrong DEFVAL for `cursor_set_column` and `cursor_set_line` in TextEdit.
- `Script::get_member_line()` now returns -1 ("found nothing") by default.
Now project files don't have to be named "godot.cfg" anymore, they can have any name so as long as it ends with *.godot.
Also godot will automatically start the editor now if launched with a project file as an argument.
This allows for double-clicking of projects to open them :)
Code-wise this should be complete, but there's still work to do:
- Make a nice icon for godot projects.
- Work on installers/packaging -> register the extension and icon with godot.
- Update the 2.1 to 3.0 exporter.
Tested on linux and windows so far.
There was a logic error in #7815 which made
Variant.hash_compare() == Variant.hash_compare() always true.
In an attempt to short-circuit the NaN check I made an (in hindsight) obvious
error: 10 == 12 || is_nan(10) == is_nan(12)
This will be true for all inputs, except for the NaN, not-NaN case. The macro
has been updated to now generate:
(10 == 12) || (is_nan(10) && is_nan(10))
so:
(10 == 12) || (is_nan(10) && is_nan(12)) = false
False or (False and False) is False
(10 == 10) || (is_nan(10) && is_nan(10)) = true
True or (False and False) is True
(Nan == 10) || (is_nan(NaN) && is_nan(10)) = false
False or (True and False) is False
(Nan == Nan) || (is_nan(NaN) && is_nan(NaN)) = true
False or (True and True) is True
Which is correct for all cases.
This bug was triggered because the hash function for floating point numbers
can very easily generate collisions for the tested Vector3(). I've also added
an extra hashing step to the float hash function to make this less likely to
occur.
This fixes#8081 and probably many more random weirdness.
Added set_scale, set_rotation_euler, set_rotation_axis_angle. Addresses #2565 directly.
Added an euler angle constructor for Basis in GDScript and also exposed is_normalized for vectors and quaternions.
Various other changes mostly cosmetic in nature.
This module was written by bojidar-bg and me, with the help of ClikCode and touilleMan.
This adds a module to Godot that enables the use of dynamic libraries as a source for scripts.
That also allows third party libraries to be linked to Godot more easily and without creating modules.
For a readme see https://github.com/GodotNativeTools/godot_headers/blob/master/README.md
This is in alignment with other functions in vector classes.
Also added checks for normalization, fixed the sign of reflect (which now corresponds to reflection along a plane mathematically), added bounce method and updated docs.
Fixes#8201.
When performing polar decomposition in 2D as B = R.S, where R is rotation (with determinant +1) and S is scaling, use the convention that reflections are absorbed into S through a reflection around y axis.
In 3D, this is done by using a reflection along all three axes, but since the dimensionality is even in 2D, one axis needs to be chosen.
Fixes Matrix32::get_rotation and Matrix32::get_scale (which weren't properly fixed in #7445).
**Breaking change**
Removed the `JOY_SNES_*` and `JOY_SEGA_*` constants. Imho there's no reason for a modern game engine to provide button aliases for decades-old hardware.
Also renamed `JOY_ANALOG_{0,1}_{X,Y}` to `JOY_ANALOG_{L,R}{X,Y}` and removed `JOY_ANALOG_2_*`.
Enabled by default as in Blender, but can be disabled separately for 2D & 3D;
the core functionality is in Input so this could be reused or even exposed to scripts in the future
When using get_tree().input_event(ev), the engine will JUST send the event down the SceneTree.
However, you won't get any of the benefits of the Input singleton:
- No InputMap actions will be emitted
- The internal input state won't be modified, so methods like `Input.get_mouse_pos()` or `Input.is_joy_button_pressed` won't return the expected output after sending the event.
This is fixed by using `Input.parse_input_event(ev)` instead.
I guess we'll also have to update the docs to reflect that this is the preferred method of sending custom InputEvents.
I can show you the code
Pretty, with proper whitespace
Tell me, coder, now when did
You last write readable code?
I can open your eyes
Make you see your bad indent
Force you to respect the style
The core devs agreed upon
A whole new world
A new fantastic code format
A de facto standard
With some sugar
Enforced with clang-format
A whole new world
A dazzling style we all dreamed of
And when we read it through
It's crystal clear
That now we're in a whole new world of code
Done:
- X11, server (tested)
- Windows (developed, would be nice to retest)
- OSX (not tested)
Prepared (not developed):
- Android (code is here, but may not compile)
- iphone
- winrt
- bb10
- haiku
- javascript
- Add FIXME tags comments to some unfixed potential bugs
- Remove some checks (always false: unsigned never < 0)
- Fix some if statements based on reviews.
- Bunch of missing `else` statements
This fixes HashMap where a key or part of a key is a floating point
number. To fix this the following has been done:
* HashMap now takes an extra template argument Comparator. This class
gets used to compare keys. The default Comperator now works correctly
for common types and floating point numbets.
* Variant implements ::hash_compare() now. This function implements
nan-safe comparison for all types with components that contain floating
point numbers.
* Variant now has a VariantComparator which uses Variant::hash_compare()
safely compare floating point components of variant's types.
* The hash functions for floating point numbers will now normalize NaN
values so that all floating point numbers that are NaN hash to the same
value.
C++ module writers that want to use HashMap internally in their modules
can now also safeguard against this crash by defining their on
Comperator class that safely compares their types.
GDScript users, or writers of modules that don't use HashMap internally
in their modules don't need to do anything.
This fixes#7354 and fixes#6947.
After discussing this with Reduz this seemed like the best way to
fix#7354. This will make composite values that contain NaN in the same
places as well as the same other values compare as the same.
Additionally non-composite values now also compare equal if they are
both NaN. This breaks IEEE specifications but this is probably what most
users expect. There is a GDScript function check for NaN if the user
needs this information.
This fixes#7354 and probably also fixes#6947
This means, if you press "F" while holding "shift" and there is and
action registered for "F" that action should be pressed.
This commit restore this behaviour, lost when implementing
is_action_just_pressed.
If you want "blocking modifiers" you should code it via script.
Fixes 6826
- TCP:
- `listen` bind to wildcard "*" -> dual stack socket
- `listen` bind to address -> socket from address type
- `connect` -> resolve using best protocol (UNSPEC), socket from address type
- UDP:
- `listen` bind to wildcard "*" -> dual stack socket
- `listen` bind to address -> socket from address type
- `put_packet`/`put_var` -> resolve using TYPE_ANY (UNSPEC), socket from address type
(to change socket type you must first call `close` it)
Serialize dictionaries adding newlines between key-value pairs
Serialize group lists also with newlines in between
Serialize string properties escaping only " and \ (needed for a good diff experience with built-in scripts and shaders)
Bonus:
Make AnimationPlayer serialize its blend times always sorted so their order is predictable in the .tscn file.
This PR is back-compat; won't break the load of existing files.
clang-format does not handle that well *at all*.
For the reference, found the relevant pieces of code with:
`ag "=[ "$'\t'"]?"$'\n'"[ "$'\t'"]?{" --ignore=thirdparty`
Made sure files in core/ and tools/ have a proper Godot license header
when written by us. Also renamed aabb.{cpp,h} and object_type_db.{cpp,h}
to rect3.{cpp,h} and class_db.{cpp,h} respectively.
Also added a proper header to core/io/base64.{c,h} after clarifying
the licensing with the original author (public domain).
When clang-format sorts includes alphabetically, MethodBind breaks.
Trying to move the object.h include upward in method_bind.h does
not seem to fix the problem, this needs investigation.
It outputs a single Dictionary with all relevant information as
keys, that will less bloat the documentation and provide all details
in one function call.
This is the follow up for the 2D changes mentioned in PR #6865. It fixes various mistakes regarding the order of matrix indices, order of transformation operations, usage of atan2 function and ensures that the sense of rotation is compatible with a left-handed coordinate system with Y-axis pointing down (which flips the sense of rotations along the z-axis). Also replaced float with real_t, and tried to make use of Matrix32 methods rather than accessing its elements directly.
Affected code in the Godot code base is also fixed in this commit.
The user code using functions involving angles such as atan2, angle_to, get_rotation, set_rotation will need to be updated to conform with the new behavior. Furthermore, the sign of the rotation angles in existing 2D scene files need to be flipped as well.
This means that each time this scene is instanced, the resource will be unique!
As such, thanks to this, the following features were implemented:
-ButtonGroup is no longer a control, it's now a resource local to the scene
-ViewportTexture can be created from the editor and set to any object, making ViewportSprite and other kind of nodes obsolete!
-Added a NOTIFICATION_TRANSLATION_CHANGED for controls that need custom code
-Sorry, editor will not update automatically because it uses a different translatio method.
This is a part of the breaking changes proposed in PR #6865, solving the issue regarding the order of affine transformations described in #2565. This PR also fixes the affected code within Godot codebase. Includes improvements to documentation too.
Another change is, Matrix3::get_scale() will now return negative scaling when the determinant of the matrix is negative. The rationale behind this is simple: when performing a polar decomposition on a basis matrix M = R.S, we have to ensure that the determinant of R is +1, such that it is a proper rotation matrix (with no reflections) which can be represented by Euler angles or a quaternion.
Also replaced the few instances of float with real_t in Matrix3 and Transform.
Furthermore, this PR fixes an issue introduced due to the API breakage in #6865. Namely Matrix3::get_euler() now only works with proper rotation matrices. As a result, the code that wants to get the rotation portion of a transform needs to use Matrix3::get_rotation() introduced in this commit, which complements Matrix3::get_scaled(), providing both parts of the polar decomposition.
Finally, it is now possible to construct a rotation matrix from Euler angles using the new constructor Matrix3::Matrix3(const Vector3 &p_euler).
-Changed SectionedPropertyEditor to support this
-Renamed Globals singleton to GlobalConfig, makes more sense.
-Changed the logic behind persisten global settings, instead of the persist checkbox, a revert button is now available
Furthermore, functions which expect a rotation matrix will now give an error simply, rather than trying to orthonormalize such matrices. The documentation for such functions has be updated accordingly.
This commit breaks code using 3D rotations, and is a part of the breaking changes in 2.1 -> 3.0 transition. The code affected within Godot code base is fixed in this commit.
That year should bring the long-awaited OpenGL ES 3.0 compatible renderer
with state-of-the-art rendering techniques tuned to work as low as middle
end handheld devices - without compromising with the possibilities given
for higher end desktop games of course. Great times ahead for the Godot
community and the gamers that will play our games!
