Install splime

Install from PyPI. Start locally.

SPLime 0.4.9 ships as one PyPI package. Install it with Python 3.13, then start the local daemon. Linux and macOS are the verified worker platforms in this release; Windows supports client and import verification.

Install SPLime View PyPI release Run as a service

Install

Available from PyPI.

The distribution is named splime; the Python import package is spl, and the daemon command is spl-daemon. Requires Python 3.13+.

Linux or macOS

python3.13 -m pip install "splime==0.4.9"
spl-daemon --version

Windows client and import verification

py -3.13 -m pip install "splime==0.4.9"
py -3.13 -m pip show splime

Isolated worker CLI on Linux or macOS

# pipx keeps the spl-daemon command in its own venv
pipx install "splime==0.4.9"
# or with uv:
uv tool install "splime==0.4.9"

spl-daemon --version

Start

Start a local worker and verify it.

The daemon stores your objects, builds environments, and runs workers. Run it in the foreground on Linux or macOS while you try things out; it listens on http://127.0.0.1:8765 by default.

Foreground run

spl-daemon --version
spl-daemon serve            # http://127.0.0.1:8765
# in another shell:
spl-daemon health
spl-daemon doctor

System service

Keep a Linux worker alive across reboots.

Let the OS supervisor own the lifecycle (start on boot, restart on crash). The inline user-unit template pins the daemon to 127.0.0.1:8765; replace %h/.local/bin/spl-daemon with the absolute path reported by command -v spl-daemon when your installer uses another location.

Linux - systemd (per-user)

# Minimal user unit (~/.config/systemd/user/spl-daemon.service)
[Unit]
Description=splime local daemon
[Service]
ExecStart=%h/.local/bin/spl-daemon serve --no-auto-port
Restart=on-failure
[Install]
WantedBy=default.target

Linux - enable it

systemctl --user daemon-reload
systemctl --user enable --now spl-daemon
loginctl enable-linger "$USER"     # survive logout / reboot
systemctl --user status spl-daemon

macOS worker macOS is a verified worker platform in 0.4.9. This release documents foreground startup only; configure your own supervisor if you need a background service.

Windows client Windows supports client and import verification in 0.4.9. A Windows worker service is not listed as a verified deployment path in this release.

Container deployment

The container image is published separately.

There is currently no verified 0.4.9 image on Docker Hub. Use the PyPI installation above for this release; do not substitute latest or an older image. Container commands will return here after the versioned image is published and verified.

View on Docker Hub →

Release status

The framework and daemon are available now from the single splime==0.4.9 PyPI package. No additional public-worker package is required.

First run

Publish and call locally before any server sync.

The daemon keeps its own registry, run state, and environment cache on the worker. Use local_only=True for a first trusted object that stays on the machine; drop it after pairing the daemon to sync into a server library.

New here? Download the full all-in-one cookbook notebook (local pipelines, publishing, NodeRemote, adapters, shared libraries) and run it top to bottom.

Local-only publish

from spl import SPLClient

client = SPLClient()
client.register_env()
client.publish(
    score_order,
    name="score_order",
    library="local-sandbox",
    create=True,
    library_display_name="Local Sandbox",
    local_only=True,
)

result = client.call("score_order", kwargs={"order_id": 42})
client.objects(scope="local", compact=True)

Upgrade

Refresh a worker without deleting daemon state.

Upgrading the package never touches the daemon home directory (registry, runs, and environment cache). Upgrade, then restart the service so the new spl-daemon takes over.

PyPI package

python3.13 -m pip install --upgrade "splime==0.4.9"
# Windows client/import verification:
py -3.13 -m pip install --upgrade "splime==0.4.9"

Restart the service

# Linux (systemd user)
systemctl --user restart spl-daemon
# Otherwise restart the foreground process or your configured supervisor.
# verify
spl-daemon --version
spl-daemon health
spl-daemon doctor