chore: netcode for entities 7.0.0 as a dependency by NoelStephensUnity · Pull Request #4158 · Unity-Technologies/com.unity.netcode.gameobjects

NoelStephensUnity · 2026-09-08T16:45:11Z

Parts of this are taken from #4156.

Purpose of this PR

N4E becomes a hard dependency at 7.0.0, and com.unity.transport moves to 6.7.0 to match it. We update our CI to match the actual minimal supported editor.
We had to disable the serialization tests that would otherwise fail under CoreCLR in the editor later (as we don't have CI for it yet here)

N4E 7.0.0 ships its unified API - GhostObject, GhostBehaviour, GhostField behind NETCODE_GAMEOBJECT_BRIDGE_EXPERIMENTAL.
testproject and minimalproject were updated to 6000.7.0b1

Jira ticket

MTT-14623

Changelog

Added: Netcode for Entities (com.unity.netcode 7.0.0) is now a hard dependency, so installing Netcode for GameObjects also brings it, and its own dependencies (Entities, Burst, Collections), into the project.

Documentation

No documentation changes or additions were necessary.

Testing & QA (How your changes can be verified during release Playtest)

Green Ci should validate this properly

Up-port

No up-port is required.

Backports

No back-port is required.

… a define N4E becomes a hard dependency at 7.0.0, and com.unity.transport moves to 6.5.0 to match it. N4E 7.0.0 ships its unified API - GhostObject, GhostBehaviour, GhostField - behind NETCODE_GAMEOBJECT_BRIDGE_EXPERIMENTAL, which leaves those types internal in the configuration NGO ships. A public type cannot derive from an internal one, so the unified path can no longer key off the package being present: the UNIFIED_NETCODE versionDefine is removed from the four assembly definitions and the symbol is set by the project instead, alongside N4E's own define. NetworkObjectBridge is internal for the same reason, and UnifiedBootstrap raises an #error when only one of the two defines is set. With N4E present the two packages share the Unity.Netcode namespace, which breaks three things outside the unified gates: - N4E's Netcode facade type shadows the namespace, so 22 XML doc crefs and one nameof that qualified with Netcode. stopped resolving. - N4E declares an INetworkStreamDriverConstructor in the shared root that the enclosing namespace finds ahead of NGO's, so the one implementation in the editor tests is fully qualified. - N4E's source generator errors on an assembly that does not reference N4E, so the editor test assembly now references it. Multiplayer Services 2.3.1 does not compile against N4E 7.0.0 and no fix is published, so it is dropped from the test project manifests. Validation moves to trunk, the only editor that has dropped N4E as a core package and can resolve the 7.0.0 request rather than its bundled 6.7.0. The unified work is preserved and still compiles with both defines set: the N4E 6.7.0 spellings are updated, and GhostObject.ApplyPostTransformMatrixScale, which N4E removed in its scale rework, is reimplemented over the new PostTransformMatrix model. The unified CI job sets the defines itself and is no longer a PR check. Also disables the serialization tests that fail under CoreCLR so the package is not ejected from the editor manifest batch. Parts of this are taken from #4156.

u-pr

💡 Harness Review

The change updates the package dependency surface and isolates the optional unified bridge behind project-level defines; I examined the runtime namespace/API migration, scale bridge call paths, assembly references, manifests, and CI gating.

_{Reviewed commit c78f679}

_{🤖 Helpful? 👍/👎}

Three things the standards job caught, since it opens the project, runs the formatter with --fix and then fails on any diff at all: - The fully qualified INetworkStreamDriverConstructor is simplified by the formatter (IDE0001), so write it the way the formatter wants. It still resolves to NGO's, since the lookup walks out to Unity.Netcode.Transports before reaching the shared root. - N4E 7.0.0 renamed the serialized GlobalNetCodeConfig field to GlobalNetcodeConfig, which the editor rewrites in the test project's settings asset the first time it loads it. - validation_editors.minimal is back on a released editor. The standards job opens the project with it, and trunk rewrites ProjectVersion.txt out from under the cleanliness check. Nothing there needs N4E present: a released editor bundles 6.7.0 as a built-in package, so the 7.0.0 request is dropped silently and the project builds with no N4E at all - verified on 6000.7.0a6, which resolves and compiles clean with N4E absent from packages-lock, PackageCache and ScriptAssemblies alike.

The formatter and the compiler want opposite things here. With N4E present the bare INetworkStreamDriverConstructor binds to N4E's copy in the shared Unity.Netcode root, which the enclosing namespace reaches before the using directive, and the class then fails to implement it. With N4E absent - which is what the standards job's editor resolves - any qualification is redundant and IDE0001 strips it back off, so the last two attempts each traded one failure for the other. A file-scope using alias satisfies both. Its target cannot be shortened, since at file scope there is no enclosing namespace to shorten it against; the use site is a single identifier with no qualified form; and IDE0005 cannot drop the alias because removing it stops the name resolving at all.

codecov-github-com · 2026-09-08T19:41:55Z

Codecov Report

Attention: Patch coverage is 0% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...cts/Runtime/SceneManagement/NetworkSceneManager.cs	0.00%	1 Missing ⚠️

@@                Coverage Diff                @@
##           develop-3.x.x    #4158      +/-   ##
=================================================
+ Coverage          78.19%   78.26%   +0.06%     
=================================================
  Files                153      154       +1     
  Lines              26272    26543     +271     
=================================================
+ Hits               20544    20774     +230     
- Misses              5728     5769      +41

Flag	Coverage Δ
NGOv2_project_testproject_ubuntu_pinnedTrunk	`76.70% <0.00%> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
.../Runtime/Components/AnticipatedNetworkTransform.cs	`80.41% <ø> (ø)`
...tcode.gameobjects/Runtime/Core/NetworkBehaviour.cs	`87.35% <ø> (+1.12%)`	⬆️
...netcode.gameobjects/Runtime/Core/NetworkManager.cs	`79.97% <ø> (ø)`
....netcode.gameobjects/Runtime/Core/NetworkObject.cs	`75.31% <ø> (ø)`
...time/NetworkVariable/AnticipatedNetworkVariable.cs	`76.64% <ø> (ø)`
...meobjects/Runtime/Spawning/NetworkPrefabHandler.cs	`66.66% <ø> (ø)`
...cts/Runtime/SceneManagement/NetworkSceneManager.cs	`79.49% <0.00%> (ø)`

... and 4 files with indirect coverage changes

Components	Coverage Δ
com.unity.netcode.gameobjects	`78.26% <0.00%> (+0.06%)`	⬆️

ℹ️ Need help interpreting these results?

…r version The 7.0.0 dependency could not be satisfied anywhere. Every editor available today - 6000.7.0a6 and current trunk alike - lists com.unity.netcode 6.7.0 as mustBeBundled in its manifest, so UPM drops a 7.0.0 request silently: no assemblies, no packages-lock entry, no warning. upm-pvp then fails hard with "depends on com.unity.netcode, which is not in the lock file", which is what took out Validate on all three platforms and Package Test on both editors. Dependency versions are minimums, so asking for 6.7.0 resolves the bundled package today and picks up 7.0.0 by itself once an editor manifest carries it. The transport bump goes back to 2.6.0: 6.5.0 was never published (the 6.x line is editor-bundled only, latest published is 2.7.4), so it fails PVP-160-1, which the Validate job explicitly requires - and N4E asks for 6.5.0 itself, so resolution is unaffected either way. With the dependency resolving everywhere, nothing needs a trunk editor and validation_editors goes back to what develop had. Since either version can now be what resolves, the unified path compiles against both, gated on a UNIFIED_NETCODE_7_0_0 versionDefine keyed to the resolved package version. The 6.7.0 spellings are aliased to the 7.0.0 names so the use sites read the same either way, except N4E's Netcode class: inside namespace Unity.Netcode that name binds to the enclosing namespace before any file-scope alias is considered, so it is aliased as EntitiesNetcode in both. GhostObjectScaleExtensions is 7.0.0-only, since 6.7.0 still has the method it stands in for. That also makes the dependency real rather than dropped, which has consequences worth knowing: N4E's assemblies now build in every project, its ClientServerBootstrap becomes the Entities bootstrap with a non-zero AutoConnectPort, and it writes a NetcodeConfig into the project when its settings asset does not resolve one - which is why the GlobalNetCodeConfig field goes back to the 6.7.0 spelling here. Verified on 6000.7.0a6 against both versions: shipping (unified off), unified against bundled 6.7.0, and unified against 7.0.0 all compile with zero errors, and the standards check is clean.

u-pr · 2026-09-10T08:38:25Z

⚠️ Some settings were ignored in '.pr_agent.toml (repository root)' repo settings

Check the configuration's TOML syntax, permitted settings, and size limit. See the error below.

Error message:
Ignored prohibited settings: github_app.handle_pr_actions. Permitted settings were applied.

Configuration content:

# This is the configuration for the u-pr bot
# https://github.cds.internal.unity3d.com/unity/u-pr
# For configuration of this file:
# https://developer.portal.internal.unity.com/catalog/default/component/u-pr/docs/configuration/automatic_runs/

[github_app]
handle_pr_actions = ['opened', 'ready_for_review']  # PR events that auto-run pr_commands
pr_commands       = ["/harness-review"]             # what runs on those events
reviewer_commands = ["/harness-review"]             # what runs when @u-pr is added as reviewer

…Technologies/com.unity.netcode.gameobjects into chore/n4e-hard-dependency

…e with N4E

N4E 7.0.0 renamed NetCodeConfig to NetcodeConfig and moved its types into Unity.Netcode. Use the 7.0.0 spelling, aliased for 6.7.0 the way the rest of the unified code already is. The editor test assembly still declared UNIFIED_NETCODE as a versionDefine after #4158 made it project-level, so its unified block compiled while the runtime's did not. It also referenced Unity.NetCode by name alongside the GUID for the same assembly. Both removed, UNIFIED_NETCODE_7_0_0 added. The snapshot packet size drops from 15000 to 4096. A snapshot is sent unreliably, so losing any one of its fragments loses the whole snapshot: 4096 spans 3 datagrams at the 1400 byte MTU where 15000 spanned 11, and still caps ~840 ghosts at the measured 4.87 bytes each. The defaults version is bumped so a project that already recorded the marker picks the new value up. Review feedback: - The experimental opt-in is gone. N4E ships as a normal 3.0.0 release. - ResolveGlobalConfig's project scan is gone. N4E assigns NetcodeConfig.Global from an InitializeOnLoadMethod and delayCall runs after those complete, so there is nothing left for a scan to catch. - The apply log uses the contextual logger at Developer level. - UnifiedIsConfiguredCorrectly reports which setting is wrong rather than discarding the reason it already computed.

* feat: auto-configure NetcodeConfig for hybrid mode When Netcode for Entities is installed and a registered network prefab carries a GhostObject, the project's NetCodeConfig is now aligned with what NGO needs rather than leaving it to the user to discover. The two settings hybrid mode cannot run without (single world hosting, and automatic bootstrapping disabled since NetworkManager owns world creation) are corrected whenever they drift. The Netcode for Entities tick rates are driven from NetworkConfig.TickRate so that ghost transform updates land on the same interval as everything else. The snapshot, interpolation and transport values tuned against the 2000 instance stress test are applied once and then left alone, so a user's own edits survive; Project Settings > Multiplayer > Netcode for GameObjects can restore them. The config is never created here. Netcode for Entities already creates one unconditionally from its own InitializeOnLoadMethod, and creating a second lands the project in its multiple-config error path. Also: - UnifiedIsConfiguredCorrectly now validates EnableClientServerBootstrap and warns when the two tick rates diverge. Its two log messages were missing their string interpolation prefixes. - m_TempStreamSize returns to the Netcode for Entities default of 8192. GhostSendSystem takes max(TempStreamInitialSize, dataStream.Capacity), and capacity is DefaultSnapshotPacketSize, so 4192 had no effect. - Unity.Netcode.Editor.Tests gains the UNIFIED_NETCODE version define. Without it any hybrid editor test compiles away and reports zero cases rather than failing. - Adds a measurement fixture for sizing DefaultSnapshotPacketSize. It is marked Explicit so its 24 cases stay out of the suites. * style Some comment clean up. * fix and test Fixing some things u-pr caught and adding a test to validate the update. * refactor Refactoring when the NetcodeConfig default settings are written. Assuring that the NetcodeConfig tick rate is always the currently set tick rate when starting a session. * test Updating to use a more NGO like testing pattern. Some comment clean up is included. * update resolving concern over the .EndVertical(); * style removing white space. * chore: n4e 7.0.0 spellings, review feedback and snapshot size N4E 7.0.0 renamed NetCodeConfig to NetcodeConfig and moved its types into Unity.Netcode. Use the 7.0.0 spelling, aliased for 6.7.0 the way the rest of the unified code already is. The editor test assembly still declared UNIFIED_NETCODE as a versionDefine after #4158 made it project-level, so its unified block compiled while the runtime's did not. It also referenced Unity.NetCode by name alongside the GUID for the same assembly. Both removed, UNIFIED_NETCODE_7_0_0 added. The snapshot packet size drops from 15000 to 4096. A snapshot is sent unreliably, so losing any one of its fragments loses the whole snapshot: 4096 spans 3 datagrams at the 1400 byte MTU where 15000 spanned 11, and still caps ~840 ghosts at the measured 4.87 bytes each. The defaults version is bumped so a project that already recorded the marker picks the new value up. Review feedback: - The experimental opt-in is gone. N4E ships as a normal 3.0.0 release. - ResolveGlobalConfig's project scan is gone. N4E assigns NetcodeConfig.Global from an InitializeOnLoadMethod and delayCall runs after those complete, so there is nothing left for a scan to catch. - The apply log uses the contextual logger at Developer level. - UnifiedIsConfiguredCorrectly reports which setting is wrong rather than discarding the reason it already computed. * feat: hybrid NetcodeConfig values applied at session start Netcode for GameObjects no longer writes anything into the NetcodeConfig asset. NetworkManager writes the in-memory global config immediately before the world is created from it, and Netcode for Entities seeds its world singletons from that. Two values are driven, both measured rather than assumed: - The tick rates. NGO's own send queues are flushed by a system in N4E's SimulationSystemGroup, so a simulation rate below NetworkConfig.TickRate starves NGO's outbound traffic. N4E defaults to 60 and NGO to 30. - The snapshot packet size, behind the new NetworkConfig.AutoConfigureSnapshotSize. N4E's default of one MTU round-robins above roughly 230 GhostObjects: at 512 instances it holds a 13.51Hz effective update rate against a 30Hz tick, where 4096 holds 30.00Hz with no saturated snapshots. Below that count the cap is never reached, so it costs nothing. UnifiedBootstrap no longer delegates to base.Initialize. Entities selects the most derived ICustomBootstrap, so this type is the only one it instantiates under UNIFIED_NETCODE and N4E's own Initialize is never reached on its own. Calling it only gave N4E a chance to create worlds this bootstrap would then reject, which is why EnableClientServerBootstrap had to be configured at all. It no longer does. Removed with it: the editor applier, the Project Settings section, the applied-defaults version marker, and the interpolation, despawn reservation and transport queue values. Those were tuned against a 2000 instance stress test and are either inside N4E's own recommended ranges or inert below the snapshot cap. The FailValidation and ClientDoesNotRequest hybrid test cases are removed pending a separate fix: the server schedules its disconnect reason into N4E's outgoing RPC buffer and then tears the connection down in the same call, before RpcSystem transmits it. * test: move the connection approval latch fix out of this PR The polled disconnect reason check belongs with the fix for the defect it found, not with the NetcodeConfig changes. Only the removal of the FailValidation hybrid fixture stays here. * chore: verbosity pass Drops a stray blank line that was the only remaining change in NetcodeForGameObjectsProjectSettings, so that file leaves the diff. Trims the comments that carried mechanism or measurements the PR description already covers, keeping the reason each value or omission exists. * chore: clarify the AutoConfigureSnapshotSize tooltip --------- Co-authored-by: Unity Netcode CI <74025435+netcode-ci-service@users.noreply.github.com>

NoelStephensUnity requested review from a team and EmandM as code owners September 8, 2026 16:45

u-pr Bot reviewed Sep 8, 2026

View reviewed changes

NoelStephensUnity requested a review from michalChrobot September 8, 2026 16:58

michalChrobot reviewed Sep 8, 2026

View reviewed changes

Comment thread testproject/ProjectSettings/ProjectSettings.asset

michalChrobot approved these changes Sep 8, 2026

View reviewed changes

NoelStephensUnity enabled auto-merge (squash) September 8, 2026 17:17

NoelStephensUnity added 2 commits September 8, 2026 12:31

NoelStephensUnity disabled auto-merge September 8, 2026 19:57

NoelStephensUnity enabled auto-merge (squash) September 8, 2026 21:38

michalChrobot mentioned this pull request Sep 9, 2026

chore: Address last issues on NGO side (following on N4Ev7.0.0 release) #4156

Closed