parallelazy-vscode
VS Code extension designed to work with
parallelazy and overcome
limitations of debugpy when debugging processes started using the spawn
method instead of fork.
When a parallelazy worker starts its debugger, this extension automatically
attaches a VS Code debugpy session to that worker. Each worker can therefore
be debugged independently, without manually finding or forwarding its
debugging port.
Requirements
- VS Code 1.80 or newer
- Python with
parallelazy
installed
debugpy, installed as a dependency of parallelazy- A
debugpy launch configuration in the workspace
Install parallelazy in the same Python environment used by the debug
configuration:
python -m pip install parallelazy
For installation from source and the framework's complete documentation, see
the parallelazy repository.
Quick start
- Install and enable this extension.
- Create a VS Code
debugpy launch configuration for the Python
application that creates the parallelazy workers.
- Enable worker debugging by constructing the manager with
PluginManager(worker_pydebug=True).
- Start the application with Run and Debug.
- When a worker calls
parallelazy.utils.init_debugger(), this extension
automatically starts an attach session for that worker.
Example launch.json:
{
"version": "0.2.0",
"configurations": [
{
"name": "Python: parallelazy application",
"type": "debugpy",
"request": "launch",
"program": "${workspaceFolder}/main.py",
"console": "integratedTerminal"
}
]
}
See the integration guide for the complete worker
example, the connection flow, configuration options, and troubleshooting.
Configuration
The extension contributes these settings:
| Setting |
Default |
Description |
debugpySpawn.host |
localhost |
Host used when attaching to a worker. |
debugpySpawn.justMyCode |
false |
Whether attached sessions debug only user code. |
The extension chooses a free local IPC port when it activates. It injects that
port into launched debugpy applications as _DEBUG_SPAWN_EXT_PORT; no
manual port configuration is required.
Scope and limitations
- Automatic setup occurs for
debugpy launch configurations.
- Worker attach requests are received over localhost TCP.
- Worker debugger ports are selected dynamically by
debugpy.
- The extension does not start the Python application or install Python
packages for you.
- Start the application from a
__main__ guard, as required by Python's
spawn multiprocessing mode.
