mirror of
https://github.com/livebook-dev/livebook.git
synced 2025-09-07 21:44:36 +08:00
79e5c432b3
* Isolate evaluation in separate node for each session * Start new remote upon first evaluation and handle nodedown * Add UI for managing evaluation node, improve naming and structure * Show runtime initialization errors and some fixes * Improve standalone node initialization * Correctly handle multiple sessions connecting to the same node * Fix session tests concerning evaluation * Documentation and some refactoring * Various improvements * Configure schedulers to get to sleep immediately after evaluation * Move EvaluatorSpervisor into the Remote namespace * Fix evaluators cleanup * Add tests * Improve flash messages * Introduce remote genserver taking care of cleanup * Redefine the Runtime protocol to serve as an interface for evaluation * Cleanup operations * Use reference for communication with a standalone node * Use shortnames for distribution by default * Update node configuration and make sure epmd is running * Rename Remote to ErlDist
64 lines
2.1 KiB
Elixir
64 lines
2.1 KiB
Elixir
defprotocol LiveBook.Runtime do
|
|
@moduledoc false
|
|
|
|
# This protocol defines an interface for evaluation backends.
|
|
#
|
|
# Usually a runtime involves a set of processes responsible
|
|
# for evaluation, which could be running on a different node,
|
|
# however the protocol does not require that.
|
|
|
|
@typedoc """
|
|
A term used to identify evaluation or container.
|
|
"""
|
|
@type ref :: term()
|
|
|
|
@doc """
|
|
Sets the caller as runtime owner.
|
|
|
|
The runtime most likely has some kind of leading process,
|
|
this method starts monitoring it and returns the monitor reference,
|
|
so the caller knows if the runtime is down by listening to a :DOWN message.
|
|
"""
|
|
@spec connect(t()) :: reference()
|
|
def connect(runtime)
|
|
|
|
@doc """
|
|
Disconnects the current owner from runtime.
|
|
|
|
This should cleanup the underlying node/processes.
|
|
"""
|
|
@spec disconnect(t()) :: :ok
|
|
def disconnect(runtime)
|
|
|
|
@doc """
|
|
Asynchronously parses and evaluates the given code.
|
|
|
|
Container isolates a group of evaluations. Every evaluation can use previous
|
|
evaluation's environment and bindings, as long as they belong to the same container.
|
|
|
|
Evaluation outputs are send to the connected runtime owner.
|
|
The messages should be of the form:
|
|
|
|
* `{:evaluation_stdout, ref, string}` - output captured during evaluation
|
|
* `{:evaluation_response, ref, response}` - final result of the evaluation
|
|
"""
|
|
@spec evaluate_code(t(), String.t(), ref(), ref(), ref()) :: :ok
|
|
def evaluate_code(runtime, code, container_ref, evaluation_ref, prev_evaluation_ref \\ :initial)
|
|
|
|
@doc """
|
|
Disposes of evaluation identified by the given ref.
|
|
|
|
This should be used to cleanup resources related to old evaluation if no longer needed.
|
|
"""
|
|
@spec forget_evaluation(t(), ref(), ref()) :: :ok
|
|
def forget_evaluation(runtime, container_ref, evaluation_ref)
|
|
|
|
@doc """
|
|
Disposes of evaluation container identified by the given ref.
|
|
|
|
This should be used to cleanup resources keeping track
|
|
of the container and contained evaluations.
|
|
"""
|
|
@spec drop_container(t(), ref()) :: :ok
|
|
def drop_container(runtime, container_ref)
|
|
end
|