Run Multiple Transports Simultaneously

This guide shows how to expose a fastcs driver through multiple protocols at once.

Basic Setup

Pass a list of transports to FastCS:

from fastcs.control_system import FastCS
from fastcs.transports import (
    EpicsCATransport,
    GraphQLTransport,
    RestTransport,
)

controller = MyController()
controller.set_path(["DEVICE"])  # PV prefix for EPICS / route prefix for REST

fastcs = FastCS(
    controller,
    [
        EpicsCATransport(),
        RestTransport(),
        GraphQLTransport(),
    ]
)
fastcs.run()

All transports run concurrently, exposing the same controller API.

Choosing controller ids across transports

Each transport derives its addressing from the controller’s id (the PV prefix for EPICS, the URL prefix for REST, the top-level Query field for GraphQL), and each enforces its own charset at startup.

Transport	Allowed id charset
EPICS CA	[A-Za-z0-9_-]+, plus the 60-char PV name limit
EPICS PVA	[A-Za-z0-9_-]+, plus the 60-char PV name limit
REST	[A-Za-z0-9_-]+
Tango	[A-Za-z0-9_-]+
GraphQL	[A-Za-z_][A-Za-z0-9_]* (GraphQL Name: no hyphens, no leading digit)

If you serve the same controller through multiple transports, use the intersection — a leading letter or underscore followed by letters, digits and underscores. GraphQL is the lowest common denominator: an id like dev-01 will start an EPICS or REST transport happily but fail fast when GraphQL is added.

Available Transports

Transport	Protocol	Install Extra	Primary Use Case
EpicsCATransport	EPICS Channel Access	fastcs[epicsca]	Control system integration
EpicsPVATransport	EPICS PV Access	fastcs[epicspva]	Modern EPICS with structured data
TangoTransport	Tango	fastcs[tango]	Tango control system
RestTransport	HTTP REST	fastcs[rest]	Web applications, debugging
GraphQLTransport	GraphQL	fastcs[graphql]	Flexible queries, web clients

Install extras as needed:

pip install "fastcs[epicsca,rest,graphql]"

Transport Configuration

Each transport has its own options:

EPICS Channel Access

The PV prefix is the first segment of the controller’s path (set via controller.set_path([...]) or auto-seeded by launch() from the YAML entry’s id:).

from pathlib import Path
from fastcs.transports import (
    EpicsCATransport,
    EpicsDocsOptions,
    EpicsGUIOptions,
)

epics_ca = EpicsCATransport(
    gui=EpicsGUIOptions(
        output_dir=Path("./opis"),
        title="Device Control",
    ),
    docs=EpicsDocsOptions(
        output_dir=Path("./reference"),
    ),
)

EPICS PV Access

from fastcs.transports import EpicsPVATransport

epics_pva = EpicsPVATransport()

REST

from fastcs.transports import RestTransport
from fastcs.transports.rest import RestServerOptions

rest = RestTransport(
    rest=RestServerOptions(
        host="0.0.0.0",
        port=8080,
        log_level="info",
    )
)

GraphQL

from fastcs.transports import GraphQLTransport
from fastcs.transports.graphql import GraphQLServerOptions

graphql = GraphQLTransport(
    graphql=GraphQLServerOptions(
        host="localhost",
        port=8081,
    )
)

Tango

from fastcs.transports import TangoTransport, TangoDSROptions

tango = TangoTransport(
    tango=TangoDSROptions(
        dsr_instance="MY_SERVER_INSTANCE",
    ),
)

The Tango device name for each controller is derived from its id — {id}/{dev_class}/{dsr_instance}. The id forms the leading device-name segment, so multiple controllers in one DSR get distinct device names.

EPICS CA + PVA Together

Run both EPICS protocols simultaneously:

from pathlib import Path

from fastcs.transports import (
    EpicsCATransport,
    EpicsGUIOptions,
    EpicsPVATransport,
)

controller.set_path(["DEVICE"])

fastcs = FastCS(
    controller,
    [
        EpicsCATransport(
            gui=EpicsGUIOptions(output_dir=Path("./opis")),
        ),
        EpicsPVATransport(),
    ]
)

Both transports derive the same PV prefix from the controller’s id and expose identical PVs.

YAML Configuration

When using the launch() framework, configure transports in YAML.

See Using the Launch Framework for details.