asgardius/virtualx-engine
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
virtualx-engine/modules/multiplayer/scene_replication_interface.cpp

922 lines
35 KiB
C++
Raw Normal View History

One Copyright Update to rule them all As many open source projects have started doing it, we're removing the current year from the copyright notice, so that we don't need to bump it every year. It seems like only the first year of publication is technically relevant for copyright notices, and even that seems to be something that many companies stopped listing altogether (in a version controlled codebase, the commits are a much better source of date of publication than a hardcoded copyright statement). We also now list Godot Engine contributors first as we're collectively the current maintainers of the project, and we clarify that the "exclusive" copyright of the co-founders covers the timespan before opensourcing (their further contributions are included as part of Godot Engine contributors). Also fixed "cf." Frenchism - it's meant as "refer to / see".
2023-01-05 13:25:55 +01:00
/**************************************************************************/
/* scene_replication_interface.cpp */
/**************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* https://godotengine.org */
/**************************************************************************/
/* Copyright (c) 2014-present Godot Engine contributors (see AUTHORS.md). */
/* Copyright (c) 2007-2014 Juan Linietsky, Ariel Manzur. */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. */
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/**************************************************************************/
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
#include "scene_replication_interface.h"
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
#include "scene_multiplayer.h"
[MP] Initial replication profiler. Part of the current network profiler stack. Tracks synchronizers, incoming/outgoing state sizes, and their bandwidth usage.
2022-11-17 03:31:23 +01:00
#include "core/debugger/engine_debugger.h"
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
#include "core/io/marshalls.h"
#include "scene/main/node.h"
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
#include "scene/scene_string_names.h"
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
#define MAKE_ROOM(m_amount) \
if (packet_cache.size() < m_amount) \
packet_cache.resize(m_amount);
[MP] Initial replication profiler. Part of the current network profiler stack. Tracks synchronizers, incoming/outgoing state sizes, and their bandwidth usage.
2022-11-17 03:31:23 +01:00
#ifdef DEBUG_ENABLED
_FORCE_INLINE_ void SceneReplicationInterface::_profile_node_data(const String &p_what, ObjectID p_id, int p_size) {
if (EngineDebugger::is_profiling("multiplayer:replication")) {
Array values;
values.push_back(p_what);
values.push_back(p_id);
values.push_back(p_size);
EngineDebugger::profiler_add_frame_data("multiplayer:replication", values);
}
}
#endif
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
SceneReplicationInterface::TrackedNode &SceneReplicationInterface::_track(const ObjectID &p_id) {
if (!tracked_nodes.has(p_id)) {
tracked_nodes[p_id] = TrackedNode(p_id);
Node *node = get_id_as<Node>(p_id);
node->connect(SceneStringNames::get_singleton()->tree_exited, callable_mp(this, &SceneReplicationInterface::_untrack).bind(p_id), Node::CONNECT_ONE_SHOT);
}
return tracked_nodes[p_id];
}
void SceneReplicationInterface::_untrack(const ObjectID &p_id) {
if (!tracked_nodes.has(p_id)) {
return;
}
uint32_t net_id = tracked_nodes[p_id].net_id;
uint32_t peer = tracked_nodes[p_id].remote_peer;
tracked_nodes.erase(p_id);
// If it was spawned by a remote, remove it from the received nodes.
if (peer && peers_info.has(peer)) {
peers_info[peer].recv_nodes.erase(net_id);
}
// If we spawned or synced it, we need to remove it from any peer it was sent to.
if (net_id || peer == 0) {
for (KeyValue<int, PeerInfo> &E : peers_info) {
E.value.spawn_nodes.erase(p_id);
}
}
}
void SceneReplicationInterface::_free_remotes(const PeerInfo &p_info) {
for (const KeyValue<uint32_t, ObjectID> &E : p_info.recv_nodes) {
Node *node = tracked_nodes.has(E.value) ? get_id_as<Node>(E.value) : nullptr;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ERR_CONTINUE(!node);
Rename queue_delete => queue_free # Conflicts: # editor/plugins/tiles/tiles_editor_plugin.cpp
2022-10-24 23:07:02 +02:00
node->queue_free();
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
}
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
bool SceneReplicationInterface::_has_authority(const Node *p_node) {
return multiplayer->has_multiplayer_peer() && p_node->get_multiplayer_authority() == multiplayer->get_unique_id();
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
void SceneReplicationInterface::on_peer_change(int p_id, bool p_connected) {
if (p_connected) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
peers_info[p_id] = PeerInfo();
for (const ObjectID &oid : spawned_nodes) {
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
_update_spawn_visibility(p_id, oid);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (const ObjectID &oid : sync_nodes) {
_update_sync_visibility(p_id, get_id_as<MultiplayerSynchronizer>(oid));
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
} else {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND(!peers_info.has(p_id));
_free_remotes(peers_info[p_id]);
peers_info.erase(p_id);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
}
void SceneReplicationInterface::on_reset() {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (const KeyValue<int, PeerInfo> &E : peers_info) {
_free_remotes(E.value);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
peers_info.clear();
// Tracked nodes are cleared on deletion, here we only reset the ids so they can be later re-assigned.
for (KeyValue<ObjectID, TrackedNode> &E : tracked_nodes) {
TrackedNode &tobj = E.value;
tobj.net_id = 0;
tobj.remote_peer = 0;
}
for (const ObjectID &oid : sync_nodes) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(oid);
ERR_CONTINUE(!sync);
sync->reset();
}
last_net_id = 0;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
void SceneReplicationInterface::on_network_process() {
Convert en_GB spelling to en_US with codespell
2023-01-21 12:25:29 +01:00
// Prevent endless stalling in case of unforeseen spawn errors.
[MP] Fix nested spawning during "ready". We want our spawns to be notified after ready, but we need to notify them in the order they entered tree, so that nested spawners can be used during "ready" (instead of having to await a frame).
2023-01-14 10:24:51 +01:00
if (spawn_queue.size()) {
ERR_PRINT("An error happened during last spawn, this usually means the 'ready' signal was not emitted by the spawned node.");
for (const ObjectID &oid : spawn_queue) {
Node *node = get_id_as<Node>(oid);
ERR_CONTINUE(!node);
if (node->is_connected(SceneStringNames::get_singleton()->ready, callable_mp(this, &SceneReplicationInterface::_node_ready))) {
node->disconnect(SceneStringNames::get_singleton()->ready, callable_mp(this, &SceneReplicationInterface::_node_ready));
}
}
spawn_queue.clear();
}
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
// Process syncs.
uint64_t usec = OS::get_singleton()->get_ticks_usec();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (KeyValue<int, PeerInfo> &E : peers_info) {
const HashSet<ObjectID> to_sync = E.value.sync_nodes;
if (to_sync.is_empty()) {
continue; // Nothing to sync
}
uint16_t sync_net_time = ++E.value.last_sent_sync;
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
_send_sync(E.key, to_sync, sync_net_time, usec);
_send_delta(E.key, to_sync, usec, E.value.last_watch_usecs);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
}
Error SceneReplicationInterface::on_spawn(Object *p_obj, Variant p_config) {
Node *node = Object::cast_to<Node>(p_obj);
ERR_FAIL_COND_V(!node || p_config.get_type() != Variant::OBJECT, ERR_INVALID_PARAMETER);
MultiplayerSpawner *spawner = Object::cast_to<MultiplayerSpawner>(p_config.get_validated_object());
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(spawner, ERR_INVALID_PARAMETER);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Track node.
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
const ObjectID oid = node->get_instance_id();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
TrackedNode &tobj = _track(oid);
[MP] Fix nested spawning during "ready". We want our spawns to be notified after ready, but we need to notify them in the order they entered tree, so that nested spawners can be used during "ready" (instead of having to await a frame).
2023-01-14 10:24:51 +01:00
// Spawn state needs to be callected after "ready", but the spawn order follows "enter_tree".
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V(tobj.spawner != ObjectID(), ERR_ALREADY_IN_USE);
tobj.spawner = spawner->get_instance_id();
[MP] Fix nested spawning during "ready". We want our spawns to be notified after ready, but we need to notify them in the order they entered tree, so that nested spawners can be used during "ready" (instead of having to await a frame).
2023-01-14 10:24:51 +01:00
spawn_queue.insert(oid);
node->connect(SceneStringNames::get_singleton()->ready, callable_mp(this, &SceneReplicationInterface::_node_ready).bind(oid), Node::CONNECT_ONE_SHOT);
return OK;
}
void SceneReplicationInterface::_node_ready(const ObjectID &p_oid) {
ERR_FAIL_COND(!spawn_queue.has(p_oid)); // Bug.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
[MP] Fix nested spawning during "ready". We want our spawns to be notified after ready, but we need to notify them in the order they entered tree, so that nested spawners can be used during "ready" (instead of having to await a frame).
2023-01-14 10:24:51 +01:00
// If we are a nested spawn, we need to wait until the parent is ready.
if (p_oid != *(spawn_queue.begin())) {
return;
}
for (const ObjectID &oid : spawn_queue) {
ERR_CONTINUE(!tracked_nodes.has(oid));
TrackedNode &tobj = tracked_nodes[oid];
MultiplayerSpawner *spawner = get_id_as<MultiplayerSpawner>(tobj.spawner);
ERR_CONTINUE(!spawner);
spawned_nodes.insert(oid);
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
if (_has_authority(spawner)) {
[MP] Fix nested spawning during "ready". We want our spawns to be notified after ready, but we need to notify them in the order they entered tree, so that nested spawners can be used during "ready" (instead of having to await a frame).
2023-01-14 10:24:51 +01:00
_update_spawn_visibility(0, oid);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
}
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
[MP] Fix nested spawning during "ready". We want our spawns to be notified after ready, but we need to notify them in the order they entered tree, so that nested spawners can be used during "ready" (instead of having to await a frame).
2023-01-14 10:24:51 +01:00
spawn_queue.clear();
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
Error SceneReplicationInterface::on_despawn(Object *p_obj, Variant p_config) {
Node *node = Object::cast_to<Node>(p_obj);
ERR_FAIL_COND_V(!node || p_config.get_type() != Variant::OBJECT, ERR_INVALID_PARAMETER);
MultiplayerSpawner *spawner = Object::cast_to<MultiplayerSpawner>(p_config.get_validated_object());
ERR_FAIL_COND_V(!p_obj || !spawner, ERR_INVALID_PARAMETER);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
// Forcibly despawn to all peers that knowns me.
int len = 0;
Error err = _make_despawn_packet(node, len);
ERR_FAIL_COND_V(err != OK, ERR_BUG);
const ObjectID oid = p_obj->get_instance_id();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (const KeyValue<int, PeerInfo> &E : peers_info) {
if (!E.value.spawn_nodes.has(oid)) {
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
continue;
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
_send_raw(packet_cache.ptr(), len, E.key, true);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
// Also remove spawner tracking from the replication state.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V(!tracked_nodes.has(oid), ERR_INVALID_PARAMETER);
TrackedNode &tobj = _track(oid);
ERR_FAIL_COND_V(tobj.spawner != spawner->get_instance_id(), ERR_INVALID_PARAMETER);
tobj.spawner = ObjectID();
spawned_nodes.erase(oid);
for (KeyValue<int, PeerInfo> &E : peers_info) {
E.value.spawn_nodes.erase(oid);
}
return OK;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
Error SceneReplicationInterface::on_replication_start(Object *p_obj, Variant p_config) {
Node *node = Object::cast_to<Node>(p_obj);
ERR_FAIL_COND_V(!node || p_config.get_type() != Variant::OBJECT, ERR_INVALID_PARAMETER);
MultiplayerSynchronizer *sync = Object::cast_to<MultiplayerSynchronizer>(p_config.get_validated_object());
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(sync, ERR_INVALID_PARAMETER);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Add to synchronizer list.
TrackedNode &tobj = _track(p_obj->get_instance_id());
const ObjectID sid = sync->get_instance_id();
tobj.synchronizers.insert(sid);
sync_nodes.insert(sid);
// Update visibility.
sync->connect("visibility_changed", callable_mp(this, &SceneReplicationInterface::_visibility_changed).bind(sync->get_instance_id()));
_update_sync_visibility(0, sync);
if (pending_spawn == p_obj->get_instance_id() && sync->get_multiplayer_authority() == pending_spawn_remote) {
// Try to apply synchronizer Net ID
ERR_FAIL_COND_V_MSG(pending_sync_net_ids.is_empty(), ERR_INVALID_DATA, vformat("The MultiplayerSynchronizer at path \"%s\" is unable to process the pending spawn since it has no network ID. This might happen when changing the multiplayer authority during the \"_ready\" callback. Make sure to only change the authority of multiplayer synchronizers during \"_enter_tree\" or the \"_spawn_custom\" callback of their multiplayer spawner.", sync->get_path()));
ERR_FAIL_COND_V(!peers_info.has(pending_spawn_remote), ERR_INVALID_DATA);
Reduce and prevent unnecessary random-access to `List` Random-access access to `List` when iterating is `O(n^2)` (`O(n)` when accessing a single element) * Removed subscript operator, in favor of a more explicit `get` * Added conversion from `Iterator` to `ConstIterator` * Remade existing operations into other solutions when applicable
2024-04-15 15:18:34 +02:00
uint32_t net_id = pending_sync_net_ids.front()->get();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
pending_sync_net_ids.pop_front();
peers_info[pending_spawn_remote].recv_sync_ids[net_id] = sync->get_instance_id();
[MP] Fix remote net ID cleanup Synchronizers for spawned nodes were not correctly keeping track of the net ID assigned by the remote, preventing the replication from performing the proper cleanup. This resulted in errors being thrown when sync messages were received after despawn (which is possible due to their unreliable nature).
2024-01-14 20:13:35 +01:00
sync->set_net_id(net_id);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Try to apply spawn state (before ready).
if (pending_buffer_size > 0) {
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
ERR_FAIL_COND_V(!node || !sync->get_replication_config_ptr(), ERR_UNCONFIGURED);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
int consumed = 0;
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
const List<NodePath> props = sync->get_replication_config_ptr()->get_spawn_properties();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Vector<Variant> vars;
vars.resize(props.size());
Error err = MultiplayerAPI::decode_and_decompress_variants(vars, pending_buffer, pending_buffer_size, consumed);
ERR_FAIL_COND_V(err, err);
if (consumed > 0) {
pending_buffer += consumed;
pending_buffer_size -= consumed;
err = MultiplayerSynchronizer::set_state(props, node, vars);
ERR_FAIL_COND_V(err, err);
}
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
return OK;
}
Error SceneReplicationInterface::on_replication_stop(Object *p_obj, Variant p_config) {
Node *node = Object::cast_to<Node>(p_obj);
ERR_FAIL_COND_V(!node || p_config.get_type() != Variant::OBJECT, ERR_INVALID_PARAMETER);
MultiplayerSynchronizer *sync = Object::cast_to<MultiplayerSynchronizer>(p_config.get_validated_object());
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(sync, ERR_INVALID_PARAMETER);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
sync->disconnect("visibility_changed", callable_mp(this, &SceneReplicationInterface::_visibility_changed));
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Untrack synchronizer.
const ObjectID oid = node->get_instance_id();
const ObjectID sid = sync->get_instance_id();
ERR_FAIL_COND_V(!tracked_nodes.has(oid), ERR_INVALID_PARAMETER);
TrackedNode &tobj = _track(oid);
tobj.synchronizers.erase(sid);
sync_nodes.erase(sid);
for (KeyValue<int, PeerInfo> &E : peers_info) {
E.value.sync_nodes.erase(sid);
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
E.value.last_watch_usecs.erase(sid);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
if (sync->get_net_id()) {
E.value.recv_sync_ids.erase(sync->get_net_id());
}
}
return OK;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
void SceneReplicationInterface::_visibility_changed(int p_peer, ObjectID p_sid) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(p_sid);
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL(sync); // Bug.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Node *node = sync->get_root_node();
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL(node); // Bug.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
const ObjectID oid = node->get_instance_id();
[MP] RPC visibility. 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.
2022-11-01 13:14:35 +01:00
if (spawned_nodes.has(oid) && p_peer != multiplayer->get_unique_id()) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
_update_spawn_visibility(p_peer, oid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
_update_sync_visibility(p_peer, sync);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
[MP] RPC visibility. 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.
2022-11-01 13:14:35 +01:00
bool SceneReplicationInterface::is_rpc_visible(const ObjectID &p_oid, int p_peer) const {
if (!tracked_nodes.has(p_oid)) {
return true; // Untracked nodes are always visible to RPCs.
}
ERR_FAIL_COND_V(p_peer < 0, false);
const TrackedNode &tnode = tracked_nodes[p_oid];
if (tnode.synchronizers.is_empty()) {
return true; // No synchronizers means no visibility restrictions.
}
if (tnode.remote_peer && uint32_t(p_peer) == tnode.remote_peer) {
return true; // RPCs on spawned nodes are always visible to spawner.
} else if (spawned_nodes.has(p_oid)) {
Minor typo and docs URL fixes
2023-02-28 13:38:01 +01:00
// It's a spawned node we control, this can be fast.
[MP] RPC visibility. 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.
2022-11-01 13:14:35 +01:00
if (p_peer) {
return peers_info.has(p_peer) && peers_info[p_peer].spawn_nodes.has(p_oid);
} else {
for (const KeyValue<int, PeerInfo> &E : peers_info) {
if (!E.value.spawn_nodes.has(p_oid)) {
return false; // Not public.
}
}
return true; // All peers have this node.
}
} else {
// Cycle object synchronizers to check visibility.
for (const ObjectID &sid : tnode.synchronizers) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(sid);
ERR_CONTINUE(!sync);
// RPC visibility is composed using OR when multiple synchronizers are present.
// Note that we don't really care about authority here which may lead to unexpected
// results when using multiple synchronizers to control the same node.
if (sync->is_visible_to(p_peer)) {
return true;
}
}
return false; // Not visible.
}
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Error SceneReplicationInterface::_update_sync_visibility(int p_peer, MultiplayerSynchronizer *p_sync) {
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(p_sync, ERR_BUG);
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
if (!_has_authority(p_sync) || p_peer == multiplayer->get_unique_id()) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
return OK;
}
const ObjectID &sid = p_sync->get_instance_id();
bool is_visible = p_sync->is_visible_to(p_peer);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
if (p_peer == 0) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (KeyValue<int, PeerInfo> &E : peers_info) {
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
// Might be visible to this specific peer.
Fix MultiplayerSynchronizer peer visibility
2022-10-16 13:26:13 +02:00
bool is_visible_to_peer = is_visible || p_sync->is_visible_to(E.key);
if (is_visible_to_peer == E.value.sync_nodes.has(sid)) {
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
continue;
}
Fix MultiplayerSynchronizer peer visibility
2022-10-16 13:26:13 +02:00
if (is_visible_to_peer) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
E.value.sync_nodes.insert(sid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
} else {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
E.value.sync_nodes.erase(sid);
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
E.value.last_watch_usecs.erase(sid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
}
return OK;
} else {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V(!peers_info.has(p_peer), ERR_INVALID_PARAMETER);
if (is_visible == peers_info[p_peer].sync_nodes.has(sid)) {
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
return OK;
}
if (is_visible) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
peers_info[p_peer].sync_nodes.insert(sid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
} else {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
peers_info[p_peer].sync_nodes.erase(sid);
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
peers_info[p_peer].last_watch_usecs.erase(sid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
return OK;
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
}
Error SceneReplicationInterface::_update_spawn_visibility(int p_peer, const ObjectID &p_oid) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
const TrackedNode *tnode = tracked_nodes.getptr(p_oid);
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(tnode, ERR_BUG);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
MultiplayerSpawner *spawner = get_id_as<MultiplayerSpawner>(tnode->spawner);
Node *node = get_id_as<Node>(p_oid);
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
ERR_FAIL_NULL_V(node, ERR_BUG);
ERR_FAIL_NULL_V(spawner, ERR_BUG);
ERR_FAIL_COND_V(!_has_authority(spawner), ERR_BUG);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V(!tracked_nodes.has(p_oid), ERR_BUG);
const HashSet<ObjectID> synchronizers = tracked_nodes[p_oid].synchronizers;
bool is_visible = true;
for (const ObjectID &sid : synchronizers) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(sid);
ERR_CONTINUE(!sync);
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
if (!_has_authority(sync)) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
continue;
}
// Spawn visibility is composed using OR when multiple synchronizers are present.
if (sync->is_visible_to(p_peer)) {
is_visible = true;
break;
}
is_visible = false;
}
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
// Spawn (and despawn) when needed.
HashSet<int> to_spawn;
HashSet<int> to_despawn;
if (p_peer) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V(!peers_info.has(p_peer), ERR_INVALID_PARAMETER);
if (is_visible == peers_info[p_peer].spawn_nodes.has(p_oid)) {
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
return OK;
}
if (is_visible) {
to_spawn.insert(p_peer);
} else {
to_despawn.insert(p_peer);
}
} else {
// Check visibility for each peers.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (const KeyValue<int, PeerInfo> &E : peers_info) {
if (is_visible) {
Remove word duplicates in comments and strings, and fix casing and punctuation
2024-02-23 20:49:13 +01:00
// This is fast, since the object is visible to everyone, we don't need to check each peer.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
if (E.value.spawn_nodes.has(p_oid)) {
// Already spawned.
continue;
}
to_spawn.insert(E.key);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
} else {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Need to check visibility for each peer.
_update_spawn_visibility(E.key, p_oid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
}
}
if (to_spawn.size()) {
int len = 0;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
_make_spawn_packet(node, spawner, len);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
for (int pid : to_spawn) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_CONTINUE(!peers_info.has(pid));
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
int path_id;
[MP] Avoid unnecessary internal ref/unrefs Access the various internal components (cache/replicator) via pointer, to avoid unnecessary overhead.
2023-09-27 01:47:23 +02:00
multiplayer_cache->send_object_cache(spawner, pid, path_id);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
_send_raw(packet_cache.ptr(), len, pid, true);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
peers_info[pid].spawn_nodes.insert(p_oid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
}
if (to_despawn.size()) {
int len = 0;
_make_despawn_packet(node, len);
for (int pid : to_despawn) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_CONTINUE(!peers_info.has(pid));
peers_info[pid].spawn_nodes.erase(p_oid);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
_send_raw(packet_cache.ptr(), len, pid, true);
}
}
return OK;
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
Error SceneReplicationInterface::_send_raw(const uint8_t *p_buffer, int p_size, int p_peer, bool p_reliable) {
ERR_FAIL_COND_V(!p_buffer || p_size < 1, ERR_INVALID_PARAMETER);
[Debugger] Move most profilers to ServersDebugger. Also splits bandwidth/rpc profiler (RPCProfiler is now in SceneDebugger).
2022-02-06 02:29:08 +01:00
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
Ref<MultiplayerPeer> peer = multiplayer->get_multiplayer_peer();
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
ERR_FAIL_COND_V(peer.is_null(), ERR_UNCONFIGURED);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
peer->set_transfer_channel(0);
[Net] Modularize multiplayer, expose MultiplayerAPI to extensions. - RPC configurations are now dictionaries. - Script.get_rpc_methods renamed to Script.get_rpc_config. - Node.rpc[_id] and Callable.rpc now return an Error. - Refactor MultiplayerAPI to allow extension. - New MultiplayerAPI.rpc method with Array argument (for scripts). - Move the default MultiplayerAPI implementation to a module.
2022-07-12 23:12:42 +02:00
peer->set_transfer_mode(p_reliable ? MultiplayerPeer::TRANSFER_MODE_RELIABLE : MultiplayerPeer::TRANSFER_MODE_UNRELIABLE);
[MP] Let MultiplayerAPI handle packet relaying and peer signaling. 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.
2022-10-08 20:50:19 +02:00
return multiplayer->send_command(p_peer, p_buffer, p_size);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Error SceneReplicationInterface::_make_spawn_packet(Node *p_node, MultiplayerSpawner *p_spawner, int &r_len) {
ERR_FAIL_COND_V(!multiplayer || !p_node || !p_spawner, ERR_BUG);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
const ObjectID oid = p_node->get_instance_id();
[MP] Fix spawned nodes not working after reset Ensures that spawnable nodes (i.e. spawned nodes over which the local instance has authority) always have a network ID, since they may lose it after the multiplayer is reset (e.g. when changing the multiplayer peer).
2024-01-14 20:09:58 +01:00
TrackedNode *tnode = tracked_nodes.getptr(oid);
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(tnode, ERR_INVALID_PARAMETER);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
[MP] Fix spawned nodes not working after reset Ensures that spawnable nodes (i.e. spawned nodes over which the local instance has authority) always have a network ID, since they may lose it after the multiplayer is reset (e.g. when changing the multiplayer peer).
2024-01-14 20:09:58 +01:00
if (tnode->net_id == 0) {
// Ensure the node has an ID.
tnode->net_id = ++last_net_id;
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
uint32_t nid = tnode->net_id;
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
ERR_FAIL_COND_V(!nid, ERR_UNCONFIGURED);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
// Prepare custom arg and scene_id
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
uint8_t scene_id = p_spawner->find_spawnable_scene_index_from_object(oid);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
bool is_custom = scene_id == MultiplayerSpawner::INVALID_ID;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Variant spawn_arg = p_spawner->get_spawn_argument(oid);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
int spawn_arg_size = 0;
if (is_custom) {
Error err = MultiplayerAPI::encode_and_compress_variant(spawn_arg, nullptr, spawn_arg_size, false);
ERR_FAIL_COND_V(err, err);
}
// Prepare spawn state.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
List<NodePath> state_props;
List<uint32_t> sync_ids;
const HashSet<ObjectID> synchronizers = tnode->synchronizers;
for (const ObjectID &sid : synchronizers) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(sid);
[MP] Optimize internal authority checks We already know which MultiplayerAPI a certain Node uses, so we don't need to retrieve it via SceneTree every time.
2023-09-27 01:12:58 +02:00
if (!_has_authority(sync)) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
continue;
}
ERR_CONTINUE(!sync);
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
ERR_FAIL_NULL_V(sync->get_replication_config_ptr(), ERR_BUG);
for (const NodePath &prop : sync->get_replication_config_ptr()->get_spawn_properties()) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
state_props.push_back(prop);
}
// Ensure the synchronizer has an ID.
if (sync->get_net_id() == 0) {
sync->set_net_id(++last_net_id);
}
sync_ids.push_back(sync->get_net_id());
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
int state_size = 0;
Vector<Variant> state_vars;
Vector<const Variant *> state_varp;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
if (state_props.size()) {
Error err = MultiplayerSynchronizer::get_state(state_props, p_node, state_vars, state_varp);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ERR_FAIL_COND_V_MSG(err != OK, err, "Unable to retrieve spawn state.");
err = MultiplayerAPI::encode_and_compress_variants(state_varp.ptrw(), state_varp.size(), nullptr, state_size);
ERR_FAIL_COND_V_MSG(err != OK, err, "Unable to encode spawn state.");
}
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
// Encode scene ID, path ID, net ID, node name.
[MP] Avoid unnecessary internal ref/unrefs Access the various internal components (cache/replicator) via pointer, to avoid unnecessary overhead.
2023-09-27 01:47:23 +02:00
int path_id = multiplayer_cache->make_object_cache(p_spawner);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
CharString cname = p_node->get_name().operator String().utf8();
int nlen = encode_cstring(cname.get_data(), nullptr);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
MAKE_ROOM(1 + 1 + 4 + 4 + 4 + 4 * sync_ids.size() + 4 + nlen + (is_custom ? 4 + spawn_arg_size : 0) + state_size);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
uint8_t *ptr = packet_cache.ptrw();
[Net] Modularize multiplayer, expose MultiplayerAPI to extensions. - RPC configurations are now dictionaries. - Script.get_rpc_methods renamed to Script.get_rpc_config. - Node.rpc[_id] and Callable.rpc now return an Error. - Refactor MultiplayerAPI to allow extension. - New MultiplayerAPI.rpc method with Array argument (for scripts). - Move the default MultiplayerAPI implementation to a module.
2022-07-12 23:12:42 +02:00
ptr[0] = (uint8_t)SceneMultiplayer::NETWORK_COMMAND_SPAWN;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ptr[1] = scene_id;
int ofs = 2;
ofs += encode_uint32(path_id, &ptr[ofs]);
ofs += encode_uint32(nid, &ptr[ofs]);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ofs += encode_uint32(sync_ids.size(), &ptr[ofs]);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ofs += encode_uint32(nlen, &ptr[ofs]);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (uint32_t snid : sync_ids) {
ofs += encode_uint32(snid, &ptr[ofs]);
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ofs += encode_cstring(cname.get_data(), &ptr[ofs]);
// Write args
if (is_custom) {
ofs += encode_uint32(spawn_arg_size, &ptr[ofs]);
Error err = MultiplayerAPI::encode_and_compress_variant(spawn_arg, &ptr[ofs], spawn_arg_size, false);
ERR_FAIL_COND_V(err, err);
ofs += spawn_arg_size;
}
// Write state.
if (state_size) {
Error err = MultiplayerAPI::encode_and_compress_variants(state_varp.ptrw(), state_varp.size(), &ptr[ofs], state_size);
ERR_FAIL_COND_V(err, err);
ofs += state_size;
}
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
r_len = ofs;
return OK;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
Error SceneReplicationInterface::_make_despawn_packet(Node *p_node, int &r_len) {
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
const ObjectID oid = p_node->get_instance_id();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
const TrackedNode *tnode = tracked_nodes.getptr(oid);
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(tnode, ERR_INVALID_PARAMETER);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
MAKE_ROOM(5);
uint8_t *ptr = packet_cache.ptrw();
[Net] Modularize multiplayer, expose MultiplayerAPI to extensions. - RPC configurations are now dictionaries. - Script.get_rpc_methods renamed to Script.get_rpc_config. - Node.rpc[_id] and Callable.rpc now return an Error. - Refactor MultiplayerAPI to allow extension. - New MultiplayerAPI.rpc method with Array argument (for scripts). - Move the default MultiplayerAPI implementation to a module.
2022-07-12 23:12:42 +02:00
ptr[0] = (uint8_t)SceneMultiplayer::NETWORK_COMMAND_DESPAWN;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
int ofs = 1;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
uint32_t nid = tnode->net_id;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ofs += encode_uint32(nid, &ptr[ofs]);
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
r_len = ofs;
return OK;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
Error SceneReplicationInterface::on_spawn_receive(int p_from, const uint8_t *p_buffer, int p_buffer_len) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V_MSG(p_buffer_len < 18, ERR_INVALID_DATA, "Invalid spawn packet received");
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
int ofs = 1; // The spawn/despawn command.
uint8_t scene_id = p_buffer[ofs];
ofs += 1;
uint32_t node_target = decode_uint32(&p_buffer[ofs]);
ofs += 4;
[MP] Avoid unnecessary internal ref/unrefs Access the various internal components (cache/replicator) via pointer, to avoid unnecessary overhead.
2023-09-27 01:47:23 +02:00
MultiplayerSpawner *spawner = Object::cast_to<MultiplayerSpawner>(multiplayer_cache->get_cached_object(p_from, node_target));
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(spawner, ERR_DOES_NOT_EXIST);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ERR_FAIL_COND_V(p_from != spawner->get_multiplayer_authority(), ERR_UNAUTHORIZED);
uint32_t net_id = decode_uint32(&p_buffer[ofs]);
ofs += 4;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
uint32_t sync_len = decode_uint32(&p_buffer[ofs]);
ofs += 4;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
uint32_t name_len = decode_uint32(&p_buffer[ofs]);
ofs += 4;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ERR_FAIL_COND_V_MSG(name_len + (sync_len * 4) > uint32_t(p_buffer_len - ofs), ERR_INVALID_DATA, vformat("Invalid spawn packet size: %d, wants: %d", p_buffer_len, ofs + name_len + (sync_len * 4)));
List<uint32_t> sync_ids;
for (uint32_t i = 0; i < sync_len; i++) {
sync_ids.push_back(decode_uint32(&p_buffer[ofs]));
ofs += 4;
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ERR_FAIL_COND_V_MSG(name_len < 1, ERR_INVALID_DATA, "Zero spawn name size.");
// We need to make sure no trickery happens here, but we want to allow autogenerated ("@") node names.
const String name = String::utf8((const char *)&p_buffer[ofs], name_len);
ERR_FAIL_COND_V_MSG(name.validate_node_name() != name, ERR_INVALID_DATA, vformat("Invalid node name received: '%s'. Make sure to add nodes via 'add_child(node, true)' remotely.", name));
ofs += name_len;
// Check that we can spawn.
Node *parent = spawner->get_node_or_null(spawner->get_spawn_path());
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(parent, ERR_UNCONFIGURED);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ERR_FAIL_COND_V(parent->has_node(name), ERR_INVALID_DATA);
Node *node = nullptr;
if (scene_id == MultiplayerSpawner::INVALID_ID) {
// Custom spawn.
ERR_FAIL_COND_V(p_buffer_len - ofs < 4, ERR_INVALID_DATA);
uint32_t arg_size = decode_uint32(&p_buffer[ofs]);
ofs += 4;
ERR_FAIL_COND_V(arg_size > uint32_t(p_buffer_len - ofs), ERR_INVALID_DATA);
Variant v;
Error err = MultiplayerAPI::decode_and_decompress_variant(v, &p_buffer[ofs], arg_size, nullptr, false);
ERR_FAIL_COND_V(err != OK, err);
ofs += arg_size;
node = spawner->instantiate_custom(v);
} else {
// Scene based spawn.
node = spawner->instantiate_scene(scene_id);
}
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(node, ERR_UNAUTHORIZED);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
node->set_name(name);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Add and track remote
ERR_FAIL_COND_V(!peers_info.has(p_from), ERR_UNAVAILABLE);
ERR_FAIL_COND_V(peers_info[p_from].recv_nodes.has(net_id), ERR_ALREADY_IN_USE);
ObjectID oid = node->get_instance_id();
TrackedNode &tobj = _track(oid);
tobj.spawner = spawner->get_instance_id();
tobj.net_id = net_id;
tobj.remote_peer = p_from;
peers_info[p_from].recv_nodes[net_id] = oid;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
// The initial state will be applied during the sync config (i.e. before _ready).
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
pending_spawn = node->get_instance_id();
pending_spawn_remote = p_from;
pending_buffer_size = p_buffer_len - ofs;
pending_buffer = pending_buffer_size > 0 ? &p_buffer[ofs] : nullptr;
pending_sync_net_ids = sync_ids;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
parent->add_child(node);
document multiplayer replication classes, small changes to multiplayer spawner
2022-07-29 01:30:30 +02:00
spawner->emit_signal(SNAME("spawned"), node);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
pending_spawn = ObjectID();
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
pending_spawn_remote = 0;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
pending_buffer = nullptr;
pending_buffer_size = 0;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
if (pending_sync_net_ids.size()) {
pending_sync_net_ids.clear();
ERR_FAIL_V(ERR_INVALID_DATA); // Should have been consumed.
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
return OK;
}
Error SceneReplicationInterface::on_despawn_receive(int p_from, const uint8_t *p_buffer, int p_buffer_len) {
ERR_FAIL_COND_V_MSG(p_buffer_len < 5, ERR_INVALID_DATA, "Invalid spawn packet received");
int ofs = 1; // The spawn/despawn command.
uint32_t net_id = decode_uint32(&p_buffer[ofs]);
ofs += 4;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
// Untrack remote
ERR_FAIL_COND_V(!peers_info.has(p_from), ERR_UNAUTHORIZED);
PeerInfo &pinfo = peers_info[p_from];
ERR_FAIL_COND_V(!pinfo.recv_nodes.has(net_id), ERR_UNAUTHORIZED);
Node *node = get_id_as<Node>(pinfo.recv_nodes[net_id]);
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(node, ERR_BUG);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
pinfo.recv_nodes.erase(net_id);
document multiplayer replication classes, small changes to multiplayer spawner
2022-07-29 01:30:30 +02:00
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
const ObjectID oid = node->get_instance_id();
ERR_FAIL_COND_V(!tracked_nodes.has(oid), ERR_BUG);
MultiplayerSpawner *spawner = get_id_as<MultiplayerSpawner>(tracked_nodes[oid].spawner);
[Modules] Replace `ERR_FAIL_COND` with `ERR_FAIL_NULL` where applicable
2023-09-09 17:40:07 +02:00
ERR_FAIL_NULL_V(spawner, ERR_DOES_NOT_EXIST);
document multiplayer replication classes, small changes to multiplayer spawner
2022-07-29 01:30:30 +02:00
ERR_FAIL_COND_V(p_from != spawner->get_multiplayer_authority(), ERR_UNAUTHORIZED);
remove child on despawn instead of just queue_delete
2022-04-23 22:01:01 +02:00
if (node->get_parent() != nullptr) {
node->get_parent()->remove_child(node);
}
Rename queue_delete => queue_free # Conflicts: # editor/plugins/tiles/tiles_editor_plugin.cpp
2022-10-24 23:07:02 +02:00
node->queue_free();
document multiplayer replication classes, small changes to multiplayer spawner
2022-07-29 01:30:30 +02:00
spawner->emit_signal(SNAME("despawned"), node);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
return OK;
}
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
bool SceneReplicationInterface::_verify_synchronizer(int p_peer, MultiplayerSynchronizer *p_sync, uint32_t &r_net_id) {
r_net_id = p_sync->get_net_id();
if (r_net_id == 0 || (r_net_id & 0x80000000)) {
int path_id = 0;
[MP] Avoid unnecessary internal ref/unrefs Access the various internal components (cache/replicator) via pointer, to avoid unnecessary overhead.
2023-09-27 01:47:23 +02:00
bool verified = multiplayer_cache->send_object_cache(p_sync, p_peer, path_id);
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
ERR_FAIL_COND_V_MSG(path_id < 0, false, "This should never happen!");
if (r_net_id == 0) {
// First time path based ID.
r_net_id = path_id | 0x80000000;
p_sync->set_net_id(r_net_id | 0x80000000);
}
return verified;
}
return true;
}
MultiplayerSynchronizer *SceneReplicationInterface::_find_synchronizer(int p_peer, uint32_t p_net_id) {
MultiplayerSynchronizer *sync = nullptr;
if (p_net_id & 0x80000000) {
[MP] Avoid unnecessary internal ref/unrefs Access the various internal components (cache/replicator) via pointer, to avoid unnecessary overhead.
2023-09-27 01:47:23 +02:00
sync = Object::cast_to<MultiplayerSynchronizer>(multiplayer_cache->get_cached_object(p_peer, p_net_id & 0x7FFFFFFF));
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
} else if (peers_info[p_peer].recv_sync_ids.has(p_net_id)) {
const ObjectID &sid = peers_info[p_peer].recv_sync_ids[p_net_id];
sync = get_id_as<MultiplayerSynchronizer>(sid);
}
return sync;
}
Add const lvalue ref to container parameters
2021-08-01 21:47:20 +02:00
void SceneReplicationInterface::_send_delta(int p_peer, const HashSet<ObjectID> &p_synchronizers, uint64_t p_usec, const HashMap<ObjectID, uint64_t> &p_last_watch_usecs) {
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
MAKE_ROOM(/* header */ 1 + /* element */ 4 + 8 + 4 + delta_mtu);
uint8_t *ptr = packet_cache.ptrw();
ptr[0] = SceneMultiplayer::NETWORK_COMMAND_SYNC | (1 << SceneMultiplayer::CMD_FLAG_0_SHIFT);
int ofs = 1;
for (const ObjectID &oid : p_synchronizers) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(oid);
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
ERR_CONTINUE(!sync || !sync->get_replication_config_ptr() || !_has_authority(sync));
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
uint32_t net_id;
if (!_verify_synchronizer(p_peer, sync, net_id)) {
continue;
}
uint64_t last_usec = p_last_watch_usecs.has(oid) ? p_last_watch_usecs[oid] : 0;
uint64_t indexes;
List<Variant> delta = sync->get_delta_state(p_usec, last_usec, indexes);
if (!delta.size()) {
continue; // Nothing to update.
}
Vector<const Variant *> varp;
varp.resize(delta.size());
const Variant **vptr = varp.ptrw();
int i = 0;
for (const Variant &v : delta) {
vptr[i] = &v;
[MP] Fix crash when sending multiple delta variants.
2023-06-11 13:43:49 +02:00
i++;
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
}
int size;
Error err = MultiplayerAPI::encode_and_compress_variants(vptr, varp.size(), nullptr, size);
ERR_CONTINUE_MSG(err != OK, "Unable to encode delta state.");
ERR_CONTINUE_MSG(size > delta_mtu, vformat("Synchronizer delta bigger than MTU will not be sent (%d > %d): %s", size, delta_mtu, sync->get_path()));
if (ofs + 4 + 8 + 4 + size > delta_mtu) {
// Send what we got, and reset write.
_send_raw(packet_cache.ptr(), ofs, p_peer, true);
ofs = 1;
}
if (size) {
ofs += encode_uint32(sync->get_net_id(), &ptr[ofs]);
ofs += encode_uint64(indexes, &ptr[ofs]);
ofs += encode_uint32(size, &ptr[ofs]);
MultiplayerAPI::encode_and_compress_variants(vptr, varp.size(), &ptr[ofs], size);
ofs += size;
}
#ifdef DEBUG_ENABLED
_profile_node_data("delta_out", oid, size);
#endif
peers_info[p_peer].last_watch_usecs[oid] = p_usec;
}
if (ofs > 1) {
// Got some left over to send.
_send_raw(packet_cache.ptr(), ofs, p_peer, true);
}
}
Error SceneReplicationInterface::on_delta_receive(int p_from, const uint8_t *p_buffer, int p_buffer_len) {
int ofs = 1;
while (ofs + 4 + 8 + 4 < p_buffer_len) {
uint32_t net_id = decode_uint32(&p_buffer[ofs]);
ofs += 4;
uint64_t indexes = decode_uint64(&p_buffer[ofs]);
ofs += 8;
uint32_t size = decode_uint32(&p_buffer[ofs]);
ofs += 4;
ERR_FAIL_COND_V(size > uint32_t(p_buffer_len - ofs), ERR_INVALID_DATA);
MultiplayerSynchronizer *sync = _find_synchronizer(p_from, net_id);
Node *node = sync ? sync->get_root_node() : nullptr;
if (!sync || sync->get_multiplayer_authority() != p_from || !node) {
ofs += size;
ERR_CONTINUE_MSG(true, "Ignoring delta for non-authority or invalid synchronizer.");
}
List<NodePath> props = sync->get_delta_properties(indexes);
Replace error checks against `size` with `is_empty`
2024-01-19 13:21:39 +01:00
ERR_FAIL_COND_V(props.is_empty(), ERR_INVALID_DATA);
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
Vector<Variant> vars;
vars.resize(props.size());
int consumed = 0;
Error err = MultiplayerAPI::decode_and_decompress_variants(vars, p_buffer + ofs, size, consumed);
ERR_FAIL_COND_V(err != OK, err);
ERR_FAIL_COND_V(uint32_t(consumed) != size, ERR_INVALID_DATA);
err = MultiplayerSynchronizer::set_state(props, node, vars);
ERR_FAIL_COND_V(err != OK, err);
ofs += size;
sync->emit_signal(SNAME("delta_synchronized"));
#ifdef DEBUG_ENABLED
_profile_node_data("delta_in", sync->get_instance_id(), size);
#endif
}
return OK;
}
Add const lvalue ref to container parameters
2021-08-01 21:47:20 +02:00
void SceneReplicationInterface::_send_sync(int p_peer, const HashSet<ObjectID> &p_synchronizers, uint16_t p_sync_net_time, uint64_t p_usec) {
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
MAKE_ROOM(/* header */ 3 + /* element */ 4 + 4 + sync_mtu);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
uint8_t *ptr = packet_cache.ptrw();
[Net] Modularize multiplayer, expose MultiplayerAPI to extensions. - RPC configurations are now dictionaries. - Script.get_rpc_methods renamed to Script.get_rpc_config. - Node.rpc[_id] and Callable.rpc now return an Error. - Refactor MultiplayerAPI to allow extension. - New MultiplayerAPI.rpc method with Array argument (for scripts). - Move the default MultiplayerAPI implementation to a module.
2022-07-12 23:12:42 +02:00
ptr[0] = SceneMultiplayer::NETWORK_COMMAND_SYNC;
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
int ofs = 1;
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ofs += encode_uint16(p_sync_net_time, &ptr[1]);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
// Can only send updates for already notified nodes.
// This is a lazy implementation, we could optimize much more here with by grouping by replication config.
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
for (const ObjectID &oid : p_synchronizers) {
MultiplayerSynchronizer *sync = get_id_as<MultiplayerSynchronizer>(oid);
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
ERR_CONTINUE(!sync || !sync->get_replication_config_ptr() || !_has_authority(sync));
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
if (!sync->update_outbound_sync_time(p_usec)) {
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
continue; // nothing to sync.
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Node *node = sync->get_root_node();
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ERR_CONTINUE(!node);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
uint32_t net_id = sync->get_net_id();
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
if (!_verify_synchronizer(p_peer, sync, net_id)) {
// The path based sync is not yet confirmed, skipping.
continue;
Add peer visibility to MultiplayerSynchronizer. MultiplayerSynchronizers can now be configured to limit their visibility to a subset of the connected peers, if the synchronized node was spawned by a MultiplayerSpawner (either automatically or via custom spawn) the given node will also be despawned remotely. The replication system doesn't have the logic to handle subspawn directly, but it is possible to handle them appropriately by manually updating the visibility of the parent before changing the one of the nested spawns via the "update_visibility" function. The visibility of each MultiplayerSynchronizer can be controlled by adding or remove filters via "[add|remove]_visibility_filter(callable)". To further optimize the network code, visibility filters can be configured to be automatically updated during idle or physics frame, or set to always require manual update (via the "update_visibility" function).
2022-07-09 20:45:30 +02:00
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
int size;
Vector<Variant> vars;
Vector<const Variant *> varp;
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
const List<NodePath> props = sync->get_replication_config_ptr()->get_sync_properties();
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
Error err = MultiplayerSynchronizer::get_state(props, node, vars, varp);
ERR_CONTINUE_MSG(err != OK, "Unable to retrieve sync state.");
err = MultiplayerAPI::encode_and_compress_variants(varp.ptrw(), varp.size(), nullptr, size);
ERR_CONTINUE_MSG(err != OK, "Unable to encode sync state.");
// TODO Handle single state above MTU.
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
ERR_CONTINUE_MSG(size > sync_mtu, vformat("Node states bigger than MTU will not be sent (%d > %d): %s", size, sync_mtu, node->get_path()));
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
if (ofs + 4 + 4 + size > sync_mtu) {
// Send what we got, and reset write.
_send_raw(packet_cache.ptr(), ofs, p_peer, false);
ofs = 3;
}
if (size) {
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
ofs += encode_uint32(sync->get_net_id(), &ptr[ofs]);
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
ofs += encode_uint32(size, &ptr[ofs]);
MultiplayerAPI::encode_and_compress_variants(varp.ptrw(), varp.size(), &ptr[ofs], size);
ofs += size;
}
[MP] Initial replication profiler. Part of the current network profiler stack. Tracks synchronizers, incoming/outgoing state sizes, and their bandwidth usage.
2022-11-17 03:31:23 +01:00
#ifdef DEBUG_ENABLED
_profile_node_data("sync_out", oid, size);
#endif
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
if (ofs > 3) {
// Got some left over to send.
_send_raw(packet_cache.ptr(), ofs, p_peer, false);
}
}
Error SceneReplicationInterface::on_sync_receive(int p_from, const uint8_t *p_buffer, int p_buffer_len) {
ERR_FAIL_COND_V_MSG(p_buffer_len < 11, ERR_INVALID_DATA, "Invalid sync packet received");
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
bool is_delta = (p_buffer[0] & (1 << SceneMultiplayer::CMD_FLAG_0_SHIFT)) != 0;
if (is_delta) {
return on_delta_receive(p_from, p_buffer, p_buffer_len);
}
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
uint16_t time = decode_uint16(&p_buffer[1]);
int ofs = 3;
while (ofs + 8 < p_buffer_len) {
uint32_t net_id = decode_uint32(&p_buffer[ofs]);
ofs += 4;
uint32_t size = decode_uint32(&p_buffer[ofs]);
ofs += 4;
Fix multiplayer replication crash in on_sync_receive. A number of early continue cases applied the packet-provided `size` without validation, allowing large uint32_t values to be treated as negative offsets and leading to segfaults. Now, we validate `size` against the buffer length immediately to avoid a crash. This could be triggered by receiving sync data for a synchronizer who's root node had just been removed, since the code path that checked for unusable sync state failed to advance the offset. Thus the next read could interpret part of the payload as such an invalid `size`. Now, we properly advance the read offset in that case (and raise a better error).
2023-02-14 11:40:06 +01:00
ERR_FAIL_COND_V(size > uint32_t(p_buffer_len - ofs), ERR_INVALID_DATA);
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
MultiplayerSynchronizer *sync = _find_synchronizer(p_from, net_id);
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
if (!sync) {
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
// Not received yet.
ofs += size;
continue;
}
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
Node *node = sync->get_root_node();
if (sync->get_multiplayer_authority() != p_from || !node) {
Fix multiplayer replication crash in on_sync_receive. A number of early continue cases applied the packet-provided `size` without validation, allowing large uint32_t values to be treated as negative offsets and leading to segfaults. Now, we validate `size` against the buffer length immediately to avoid a crash. This could be triggered by receiving sync data for a synchronizer who's root node had just been removed, since the code path that checked for unusable sync state failed to advance the offset. Thus the next read could interpret part of the payload as such an invalid `size`. Now, we properly advance the read offset in that case (and raise a better error).
2023-02-14 11:40:06 +01:00
// Not valid for me.
ofs += size;
ERR_CONTINUE_MSG(true, "Ignoring sync data from non-authority or for missing node.");
[MP] Allow multiple synchronizers per node. 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.
2022-09-30 22:35:56 +02:00
}
if (!sync->update_inbound_sync_time(time)) {
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
// State is too old.
ofs += size;
continue;
}
[MP] Optimize internal SceneReplicationConfig access Use direct pointer addressing to avoid unnecessary refs/unrefs
2023-10-02 10:56:21 +02:00
const List<NodePath> props = sync->get_replication_config_ptr()->get_sync_properties();
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
Vector<Variant> vars;
vars.resize(props.size());
int consumed;
Error err = MultiplayerAPI::decode_and_decompress_variants(vars, &p_buffer[ofs], size, consumed);
ERR_FAIL_COND_V(err, err);
err = MultiplayerSynchronizer::set_state(props, node, vars);
ERR_FAIL_COND_V(err, err);
ofs += size;
[MP] Add a "synchronized" signal to MultiplayerSynchronized. Emitted upon receiving a valid sync packet after setting the variables state.
2023-02-20 14:46:39 +01:00
sync->emit_signal(SNAME("synchronized"));
[MP] Initial replication profiler. Part of the current network profiler stack. Tracks synchronizers, incoming/outgoing state sizes, and their bandwidth usage.
2022-11-17 03:31:23 +01:00
#ifdef DEBUG_ENABLED
_profile_node_data("sync_in", sync->get_instance_id(), size);
#endif
[Net] New replication interface, spawner and synchronizer nodes. Initial implementation of the MultiplayerReplicationInterface and its default implementation (SceneReplicationInterface). New MultiplayerSpawner node helps dealing with instantiation of scenes on remote peers (e.g. clients). It supports both custom spawns via a `_spawn_custom` virtual function, and optional auto-spawn of known scenes via a TypedArray<PackedScenes> property. New MultiplayerSynchornizer helps synchronizing states between the local and remote peers, supports both sync and spawn properties and is configured via a `SceneReplicationConfig` resource. It can also sync via path (i.e. without being spawned by a MultiplayerSpawner if both peers has it in tree, but will not send the spawn state in that case, only the sync one.
2021-10-08 14:13:06 +02:00
}
return OK;
}
[MP] Implement "watched" properties. Checked at "delta_interval" (default = every frame), synchronized (reliably) if changes are detected.
2023-03-28 09:30:58 +02:00
void SceneReplicationInterface::set_max_sync_packet_size(int p_size) {
ERR_FAIL_COND_MSG(p_size < 128, "Sync maximum packet size must be at least 128 bytes.");
sync_mtu = p_size;
}
int SceneReplicationInterface::get_max_sync_packet_size() const {
return sync_mtu;
}
void SceneReplicationInterface::set_max_delta_packet_size(int p_size) {
ERR_FAIL_COND_MSG(p_size < 128, "Sync maximum packet size must be at least 128 bytes.");
delta_mtu = p_size;
}
int SceneReplicationInterface::get_max_delta_packet_size() const {
return delta_mtu;
}
Reference in a new issue Copy permalink