This change aims to reduce the number of places that need to be changed
when adding or editing a Godot type to the bindings.
Since the addition of `Variant.From<T>/As<T>` and
`VariantUtils.CreateFrom<T>/ConvertTo<T>`, we can now replace a lot of
the previous code in the bindings generator and the source generators
that specify these conversions for each type manually.
The only exceptions are the generic Godot collections (`Array<T>` and
`Dictionary<TKey, TValue>`) which still use the old version, as that
one cannot be matched by our new conversion methods (limitation in the
language with generics, forcing us to use delegate pointers).
The cleanup applies to:
- Bindings generator:
- `TypeInterface.cs_variant_to_managed`
- `TypeInterface.cs_managed_to_variant`
- Source generators:
- `MarshalUtils.AppendNativeVariantToManagedExpr`
- `MarshalUtils.AppendManagedToNativeVariantExpr`
- `MarshalUtils.AppendVariantToManagedExpr`
- `MarshalUtils.AppendManagedToVariantExpr`
This commit replaces most usages of `ConvertManagedObjectToVariant` and
`ConvertVariantToManagedObjectOfType`, by using the `Godot.Variant`
struct instead of `System.Object`.
The most notable change is to the `GetGodotPropertyDefaultValues` method
that's generated for scripts. The dictionary it returns now stores
`Godot.Variant` values.
Remaining usages are:
- The `DelegateUtils` class, for the serialization of closure display
classes during assembly reloading by the editor. These display classes
are compiler generated classes to store values captured by a closure.
Since it's generated by the compiler, the only way we have to access
the fields is through reflection. This leads to using `System.Object`.
- Converting parameters when invoking constructors from the engine.
This will be replaced with source generators in the future.
- Legacy support for old `GetGodotPropertyDefaultValues` return values.
We need to keep supporting the old version of this generated method
for some time. Otherwise, if loading a project built with the previous
version, it could lead to the loss of exported property values.
Ideally, we should remove this legacy support before a stable release.
Its two usages were:
- The Array `ICollection.CopyTo` implementation.
It's possible that this class shouldn't be implementing the
non-generic `ICollection`, but this commit doesn't change that.
The new implementation stores the elements as boxed `Variant` values.
- The `Variant.Obj` property.
I'm not sure if this property's existence is justified, but for now
I rewrote it as a simpler version of `ConvertVariantToManagedObject`.
Update the docs to clarify that the `@icon` annotation does not work
when only attaching a script to a node, and clarify that the script itself
must be a globally accessible script type.
Add necessary build flags and switch from using a
GLES2 context to a GLES3 one.
This also enables building for OpenXR
Co-authored-by: m4gr3d <fhuyakou@gmail.com>
Co-authored-by: dsnopek <dsnopek@gmail.com>
- Removed `UnicodeAt`
- Removed `EndsWith`
- Removed `LPad` and `RPad`
- Deprecated `BeginsWith` in favor of `string.StartsWith`
- Deprecated `LStrip` and `RStrip` in favor of `string.TrimStart` and `string.TrimEnd`
- Remove `VariantSpanDisposer`, no need to dispose of the Variant Spans
since we are now borrowing the Variants instead of copying them.
- Remove `VariantSpanExtensions.Cleared` that was only used so the
Span was initialized for `VariantSpanDisposer` to know what to dispose.
- Fix stackalloc Spans to use constant VarArgsSpanThreshold
and avoid bound checks.
- Replaced `MD5Buffer`, `MD5Text`, `SHA256Buffer` and `SHA256Text` implementation to use the `System.Security.Cryptography` classes and avoid marshaling.
- Added `SHA1Buffer` and `SHA1Text`.
- Renamed `ToUTF8` to `ToUTF8Buffer`.
- Renamed `ToAscii` to `ToASCIIBuffer`.
- Added `ToUTF16Buffer` and `ToUTF32Buffer`.
- Added `GetStringFromUTF16` and `GetStringFromUTF32`.
- Renamed `IsValidInteger` to `IsValidInt`.
- Added `IsValidFileName`.
- Added `IsValidHexNumber`.
- Added support for IPv6 to `IsValidIPAddress`.
- Added `ValidateNodeName`.
- Updated the documentation of the `IsValid*` methods.
- Moved `GetBaseName` to keep methods alphabetically sorted.
- Removed `Length`, users should just use the Length property.
- Removed `Insert`, string already has a method with the same signature that takes precedence.
- Removed `Erase`.
- Removed `ToLower` and `ToUpper`, string already has methods with the same signature that take precedence.
- Removed `FindLast` in favor of `RFind`.
- Replaced `RFind` and `RFindN` implemenation with a ca ll to `string.LastIndexOf` to avoid marshaling.
- Added `LPad` and `RPad`.
- Added `StripEscapes`.
- Replaced `LStrip` and `RStrip` implementation with a call to `string.TrimStart` and `string.TrimEnd`.
- Added `TrimPrefix` and `TrimSuffix`.
- Renamed `OrdAt` to `UnicodeAt`.
- Added `CountN` and move the `caseSensitive` parameter of `Count` to the end.
- Added `Indent` and `Dedent`.
Embree initially only supported x86_64, then got arm64 support added.
Now it seems to be possible to build it with Emscripten (wasm32) and
on x86_32 Windows.
These callbacks are used for marshaling by callables and generic Godot
collections.
C# generics don't support specialization the way C++ templates do.
I knew NativeAOT could optimize away many type checks when the types
are known at compile time, but I didn't trust the JIT would do as good
a job, so I initially went with cached function pointers.
Well, it turns out the JIT is also very good at optimizing in this
scenario, so I'm changing the methods to do the conversion directly,
rather than returning a function pointer for the conversion.
The methods were moved to `VariantUtils`, and were renamed from
`GetFromVariantCallback/GetToVariantCallback` to `ConvertTo/CreateFrom`.
The new implementation looks like it goes through many `if` checks
at runtime to find the right branch for the type, but in practice it
works pretty much like template specialization. The JIT only generates
code for the relevant branch. Together with inlining, the result is
very close or the same as doing the conversion manually:
```cs
godot_variant variant;
int foo = variant.Int;
int bar = VariantUtils.ConvertTo<int>(variant);
```
If the type is a generic Godot collection, the conversion still goes
through a function pointer call.
The new code happens to be much shorter as well, with the file going
from 1057 lines to 407.
Side note: `Variant.cs` was mistakenly created in the wrong folder,
so I moved it to the `Core` folder.
This happens too often with normal usage of the API.
The warning can still be useful to find actual bugs where discarding the return
value wasn't intentional, but this should stay enabled manually, at least until
we either improve the API to remove false positives, or improve the warning (e.g.
to only warn about unused return value on const functions).
changed RETURN_VALUE_DISCARDED GDscript warning text to mention how the return value of a function is discarded; update GDScript parser warning test to include new warning text.
Adds a OfflineMultiplayerPeer class which behaves like a server with no
connected peers.
Use OfflineMultiplayerPeer as default for SceneMultiplayer.
This means that the SceneTree will act as the multiplayer authority by
default.
Calls to is_server will return true, and calls to get_unique_id will
return TARGET_PEER_SERVER.
Implemented using MultiplayerSynchronizers.
If you didn't use the synchronizer visibility features, nothing changes.
If you were using visibility, RPCs to broadcast should now behave as
expected in most configurations (i.e. by sending the RPC to _visible_
peers).
If you want to limit the visibility of RPCs for a node, add a
synchronizer for it, and configure the visibility via
"set_visibility_for" or by adding a visibility filter.
Fix RPC profiler and add average RPC size.
Improve bandwidth debugger to account for all multiplayer traffic
(excluding the lower level peer transformations).
The Zipfile Reference should be set to `NULL` when the `ZIPPacker` is closed not when a file in it is closed.
When calling `ZIPPacker.close` without this nothing happens because `zf` is `NULL`. (7zip could still extract the file but warned about unexpected end of file.)
Updates to volk, vulkan headers, `vk_enum_string_helper.h`, glslang,
spirv-reflect.
No update to VMA which still has 3.0.1 as it's last tagged release.
- Removed empty paragraphs in XML.
- Consistently use bold style for "Example:", on a new line.
- Fix usage of `[code]` when hyperlinks could be used (`[member]`, `[constant]`).
- Fix invalid usage of backticks for inline code in BBCode.
- Fix some American/British English spelling inconsistencies.
- Other minor fixes spotted along the way, including typo fixes with codespell.
- Don't specify `@GlobalScope` for `enum` and `constant`.
Add few methods to allow peers to exchange authentication information.
- `set_auth_callback(callback)`: Enable the authentication features.
The callback is a `Callable` that accepts an `int` (the peer ID), and
a `PackedByteArray` of data.
- The `peer_authenticating(id)` signal will be emitted instead of
`peer_connected` when a new peer connects.
- Use `send_auth(id: int, data: PackedByteArray)` to exchange data.
- Call `complete_auth(id: int)` when the authentication process is
complete and you expect to start receiving game data.
- The `peer_connected` signal will be emitted as soon as both parties
complete the authentication.
- Use `disconnect_peer(id)` to disconnect a connected peer.
- If the `peer_connected` signal didn't fire for that peer (i.e. it was
still in the authentication phase), the `peer_auth_failed` signal will
be emitted instead of `peer_disconnected`.
This check was removed because it introduces a bug which prevents
ZIPPacker from actually adding any files, since it must be opened before
adding any files (and therefore shouldn't be NULL at the start of
`start_file`).
This allows using generic Godot collections as type arguments for other
generic Godot collections. This also allows generic Godot collections
as parameter or return type in dynamic Callable invocations.
We aim to make the C# API reflection-free, mainly for concerns about
performance, and to be able to target NativeAOT in refletion-free mode,
which reduces the binary size.
One of the main usages of reflection still left was the dynamic
invokation of callable delegates, and for some time I wasn't sure
I would find an alternative solution that I'd be happy with.
The new solution uses trampoline functions to invoke the delegates:
```
static void Trampoline(object delegateObj, NativeVariantPtrArgs args, out godot_variant ret)
{
if (args.Count != 1)
throw new ArgumentException($"Callable expected 1 arguments but received {args.Count}.");
string res = ((Func<int, string>)delegateObj)(
VariantConversionCallbacks.GetToManagedCallback<int>()(args[0])
);
ret = VariantConversionCallbacks.GetToVariantCallback<string>()(res);
}
Callable.CreateWithUnsafeTrampoline((int num) => "Foo" + num, &Trampoline);
```
Of course, this is too much boilerplate for user code. To improve this,
the `Callable.From` methods were added. These are overloads that take
`Action` and `Func` delegates, which covers the most common use cases:
lambdas and method groups:
```
// Lambda
Callable.From((int num) => "Foo" + num);
// Method group
string AppendNum(int num) => "Foo" + num;
Callable.From(AppendNum);
```
Unfortunately, due to limitations in the C# language, implicit
conversions from delegates to `Callable` are not supported.
`Callable.From` does not support custom delegates. These should be
uncommon, but the Godot C# API actually uses them for event signals.
As such, the bindings generator was updated to generate trampoline
functions for event signals. It was also optimized to use `Action`
instead of a custom delegate for parameterless signals, which removes
the need for the trampoline functions for those signals.
The change to reflection-free invokation removes one of the last needs
for `ConvertVariantToManagedObjectOfType`. The only remaining usage is
from calling script constructors with parameters from the engine
(`CreateManagedForGodotObjectScriptInstance`). Once that one is made
reflection-free, `ConvertVariantToManagedObjectOfType` can be removed.
MultiplayerPeer changes:
- Adds is_server_relay_supported virtual method
Informs the upper MultiplayerAPI layer if it can signal peers connected
to the server to other clients, and perform packet relaying among them.
- Adds get_packet_channel and get_packet_mode virtual methods
Allows the MultiplayerAPI to retrieve the channel and transfer modes to
use when relaying the last received packet.
SceneMultiplayerPeer changes:
- Implement peer signaling and packet relaying when the MultiplayerPeer
advertise they are supported.
ENet, WebRTC, WebSocket changes:
- Removed custom code for relaying from WebSocket and ENet, and let it
be handled by the upper layer.
- Update WebRTC to split create_client, create_server, and create_mesh,
with the latter behaving like the old initialize with
"server_compatibility = false", and the first two supporting the upper
layer relaying protocol.
- Made use of [param] more frequently,
- Link to other classes' documentation more often, improve the examples.
- Made the writing style closer to how the rest of the documentation is formatted.
- Ensure these are called "functions", not "methods".
- Add [b]Warning:[/b] where more appropriate than [b]Note:[/b]
Most notably, removed " It must be a static string, so format strings can't be used.", as this behavior is actually a bug.
The bug was caused by not checking the TCP CONNECTING state
appropriately during the client handshake, and not checking the TCP
CONNECTED state during connection (which is unlikely, but might still
happen).
We use collectible AssemblyLoadContexts as that's the only way to allow
reloading assemblies after building. However, collectible assemblies
have some restrictions:
- https://learn.microsoft.com/en-us/dotnet/framework/reflection-and-codedom/collectible-assemblies#restrictions-on-collectible-assemblies
Those restrictions can cause issues with third-party code, such as some
mocking libraries.
In order to work around this problem, we're going to load assemblies
as collectible only in Godot editor, and not when running games.
These issues will still exist in the editor, but this will be enough
for some users.
We were compiling `src/loaders/external_png` but using the header from
`src/loaders/png`, leading to `-Wodr` warnings.
Remove unused bundled lodepng code since we use `external_png` (libpng).
Whitespace changes are from upstream, will attempt PR'ing a cleanup there for
future updates.
This commit is a huge refactor of the websocket module.
The module is really old, and some design choices had to be
re-evaluated.
The WebSocketClient and WebSocketServer classes are now gone, and
WebSocketPeer can act as either client or server.
The WebSocketMultiplayerPeer class is no longer abstract, and implements
the Multiplayer API on top of the lower level WebSocketPeer.
WebSocketPeer is now a "raw" peer, like StreamPeerTCP and StreamPeerTLS,
so it emits no signal, and just needs polling to update its internal
state.
To use it as a client, simply call WebSocketPeer.coonect_to_url, then
frequently poll the peer until STATE_OPEN is reached and then you can
write or read from it, or STATE_CLOSED and then you can check the
disconnect code and reason).
To implement a server instead, a TCPServer must be created, and the
accepted connections needs to be provided to
WebSocketPeer.accept_stream (which will perform the HTTP handshake).
A full example of a WebSocketServer using TLS will be provided in the
demo repository.
Replace all TODO uses of `#warning` by proper TODO comments, and will open
matching bug reports to keep track of them.
We don't have a great track record fixing TODOs, but I'd wager we're even
worse for fixing these "TODO #warning" so we should prohibit this usage.
The function that was supposed to implement the saving in
image_loader_jpegd was just returning OK without doing anything.
Copied the code from _jpgd_buffer_save_func to _jpgd_save_func but
changed the ImageLoaderJPGOSBufferto a ImageLoaderJPGOSFile to save
to a file instead of memory. Changed the image format from
FORMAT_ETC2_RGB8 to FORMAT_RGB8 since the first one was creating
a weird greyscale interlaced image.
- Use `long` and `double` types since signals currently only support 64-bit types.
- Fix bug for checking if the type name is a class registered in ClassDB.
Ensures that the versions always match the Godot version, albeit following
SemVer 2.0 so inserting a dot between "beta" and the build number.
For "stable" status, we omit the suffix as this would be interpreted as a
pre-release build too.
So we have:
| Godot version | Nupkg version |
| -------------- | -------------- |
| 4.0.0-beta | 4.0.0-beta |
| 4.0.0-beta2 | 4.0.0-beta.2 |
| 4.0.0-rc1 | 4.0.0-rc.1 |
| 4.0.0-stable | 4.0.0 |
- Add leading and trailing spaces within dictionaries, as the `{}`
characters are hard to distinguish from `[]` on some fonts.
This is especially helpful with empty arrays and dictionaries.
Implement `GDScriptTextDocument::willSaveWaitUntil` to clean up outdated documents before saving,
then update the document in `GDScriptTextDocument::didSave`.
The same node can now be controlled by multiple synchronizers at once.
Spawn visibility is composed in OR across multiple synchronizers (i.e.
if any synchronizer is visible to the peer, the object will be spawned).
Synchronizers controlling spawned nodes can now have a different
authority then the one spawning the node, in that case, the spawn
variables for that specific synchronizer are simply ignored (other
synchronizers with the same authority of the spawner will correctly
synchronize variables on spawn).
The SceneReplicationState class has been completly removed, since it was
bloating the implementation, the state is now simply stored in the
SceneReplicationInterface.
Android was the last platform to still attempt to disable RTTI (for binary
size), but both the Android editor and now the ICU library used by templates
need RTTI.
There could still be the possibility to support this for non-ICU template
builds (i.e. without the TextServerAdvanced module), but since this isn't one
of the build configurations we test regularly it's pretty risky to keep this
option only for that specific use case. And our code is already littered with
`dynamic_cast`s which weren't guarded with `!defined(NO_SAFE_CAST)`.
