"The internet is a series of pipes" — Some guy, probably.

NOT PRODUCTION READY. HEAVILY EXPERIMENTAL

A networked distributed-object framework for Panda3D built on top of Steam networking sockets. It provides authoritative server snapshots with delta compression, client-side interpolation, zone-based interest management, and a DC (distributed class) schema for defining replicated object fields — all driven by a tick-based simulation loop.

The original foundation of Panda3D pipes is based on the work of Brian Lach from the Panda3D player project.

• Snapshot replication — The server captures the state of every distributed object each tick and sends delta-compressed snapshots to clients. Only changed fields are transmitted.
• Client-side interpolation — Templated C++ InterpolatedVariable types (float, Vec2/3/4, Quaternion) record value history and interpolate smoothly between received snapshots using linear or Hermite curves, with configurable extrapolation.
• Zone-based interest — Clients subscribe to zones; the server only sends object data for zones a client cares about.
• DC schema — Shared .dc files define the replicated fields, ownership rules (ownsend, clsend), and broadcast behaviour for each distributed class.
• Distributed object lifecycle — Objects follow a well-defined lifecycle: Fresh → Generated → Alive → Disabled → Deleted, with separate client, server (AI), and owner views.
• Clock drift correction — A drift manager keeps the client simulation clock aligned to the server tick using sliding-window offset averaging and configurable correction thresholds.
• Ping / RTT tracking — Built-in ping protocol with a sliding-window average; per-client latency is available on both the server and client.
• Per-client rate control — Update rate, command rate, and interpolation amount are individually negotiable per client.
• C++ performance core — Snapshot packing, field unpacking (CClientRepository), frame management, packed objects, and math utilities are implemented in C++ and exposed to Python via Panda3D's interrogate system.
• Steam networking transport — Uses Valve's SteamNetworkingSockets for reliable/unreliable messaging, NAT traversal, and optional Steam authentication.

• Panda3D 1.10.14+
• panda3d-steamworks — Python/C++ bindings for Steamworks networking
• panda3d-toolbox — Runtime helpers used by the framework
• Steam client running (required for Steam networking sockets)
• CMake 3.16+, a C++ compiler, and Python 3.10+

pip install .

This invokes the CMake / interrogate pipeline via setuptools to compile the C++ extension module (panda3d_pipes).

Set SETUPTOOLS_SCM_PRETEND_VERSION to control the wheel version if building outside of CI.

See the examples/ directory for a minimal client/server demo. In two terminals from the repository root:

# Terminal 1 — start the server
python examples/server.py

# Terminal 2 — connect a client
python examples/client.py

The server listens on port 27015, spawns avatars for connecting clients, and relays chat messages. The client connects, sends a greeting, and periodically moves its avatar.

Configuration is done through Panda3D PRC variables. Key defaults:

The effective interpolation delay is max(pipes-interp, pipes-interp-ratio / pipes-updaterate).

This project is licensed under the BSD 3-Clause License — see LICENSE for details.

The Steamworks SDK is Copyright © Valve Corporation and is subject to the rules outlined at Distributing Open Source.