[Erwan MATHIEU (wawanbreton)]

This PR provides a few changes that improve the communication between Cura front-end and CuraEngine, to make sure that we can send large objects and/or build plates without crashes or visible latency:

• Replace libArcus/pyArcus by asio-grpc/grpcio
  • Use modern robust libraries instead of custom code
  • Can handle large amounts of data, super fast
  • Protobuf classes are generated at development instead of runtime (front-end only)
  • asio-grpc is already a dependency of CuraEngine for the communication with the plugins
• Split the single huge Slice message sent to the engine into multiple messages sent in a row
  • Allows sending many very large models for a single slice
• Start sending messages to the engine as soon as they are individually completed
  • Less waiting time before we actually start slicing
  • Reduces the max amount of memory used by the front-end

In practice, this seems to significantly reduce the time spent in the front-end to generate and send the slicing data. When using the dragon cup model, I went from ~25s with the UI frozen to not noticeable. I could also send a build plate containing 50 of them without issues.

Requires Ultimaker/Cura#21892
Requires Ultimaker/CuraEngine#2418
CURA-13217

[github-actions]

Test Results

2 403 tests   2 388 ✅  16s ⏱️
    1 suites     15 💤
    1 files        0 ❌

Results for commit 8f1c447.

♻️ This comment has been updated with latest results.

[Copilot]

🟡 Changes recommended

Unresolved backend regressions can crash engine-command lookup, silently discard accepted messages, and let overlapping streams corrupt each other’s connection state.

5 open issues

• Use Application instead of the removed UM import · New
• Prevent overlapping streams from sharing or clearing state · New
• Preserve queued messages for the upcoming stream · New
• Avoid recursive engine startup when no command is available · New
• Only cancel the context owned by the ending stream · New

What changed in this PR

This PR migrates Uranium’s backend communication from Arcus to gRPC, supporting the coordinated Cura and CuraEngine changes for streaming large slicing workloads with less latency and memory use.

Changes:

• Introduces a gRPC server with queued outgoing messages and concurrent incoming-message handling.
• Updates backend startup, shutdown, and message dispatch for gRPC.
• Replaces Arcus dependencies and adapts backend tests.

File	Description
UM/​Backend/​SignalSocket.py	Removes the Arcus socket wrapper.
UM/​Backend/​GrpcBackendServer.py	Adds gRPC streaming, message queues, and lifecycle handling.
UM/​Backend/​Backend.py	Integrates gRPC communication and updates engine startup.
tests/​TestBackend.py	Adapts backend tests and adds streaming and queue checks.
plugins/​Tools/​CameraTool/​tests/​TestCameraTool.py	Removes the Arcus import workaround.
conanfile.py	Removes the Arcus shared-library option.
conandata.yml	Replaces Arcus with gRPC and Protobuf dependencies and generation tools.

_{Review effort: Balanced}