- Category
- Developer Tools
- Rank
- No. 911Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- browser-use
- GitHub
- 122 stars
- Latest release
- 1.5.6
- Date
About
Production-ready Python event bus with async/sync handlers, FIFO/concurrency options, and WAL persistence. Powers the browser-use library.
What it does
Bubus coordinates work inside async Python applications through typed events and registered handlers. Events are validated as Pydantic models, can trigger child events, collect handler results, wait for matching responses, and move between linked buses while tracking their route and parentage.
Why it's ranked here
Bubus has unusual depth for an in-process event system. It combines runtime validation, typed return values, nested-event causality, configurable parallel handlers, forwarding loop prevention, timeouts, and readable execution trees. The supplied tests cover ordering, parallel execution, persistence, serialization, and failure behavior.
What's good
Nested work is a first-class concept, not an improvised callback chain. Parent and child relationships are tracked automatically, and awaiting a child inside a handler avoids deadlock by processing it immediately. Handler results remain attached to events and can be retrieved individually, grouped by handler, or flattened into lists and dictionaries.
Tradeoffs
It requires Python 3.11 or newer and dispatch needs a running event loop. Serial handler execution is the default; parallel handlers require explicit configuration. Awaited child events jump the normal queue order, so strict FIFO has a documented exception. Event history defaults to 50 entries. The library supplies no network transport between processes or services.
How to use it well
Use Bubus for typed orchestration inside an async Python process, especially when workflows spawn nested tasks, need correlated responses, or benefit from traceable handler results. Define narrow event models, set realistic timeouts, and choose serial or parallel handlers deliberately. Pair it with a message broker or RPC layer when events must cross process or service boundaries.
Technical notes+
bubus/service.py implements EventBus, a clean-shutdown asyncio queue, a process-global reentrant lock, FIFO dispatch, optional parallel handlers, bounded history, forwarding, and WAL configuration. bubus/models.py defines generic Pydantic BaseEvent and EventResult models, UUID validation, handler protocols, result typing, timestamps, route tracking, and await behavior that processes nested queued events to prevent deadlock. bubus/logging.py renders event, handler, timing, child, error, and timeout trees. bubus/helpers.py adds retry and semaphore machinery, including process-wide locking through portalocker with an asyncio fallback. tests/test_eventbus.py, tests/test_handler_timeout.py, and tests/test_semaphores.py exercise core dispatch, ordering, handler modes, nested timeouts, and multiprocess concurrency.
Observed
- License
- MIT License
- Primary language
- Python
- Runtime support
- Python 3.11 or newer, below Python 4
- Install surface
- Python package installed with pip
- Interface
- Importable Python library
- Packaging
- Hatchling build backend
- Platform support
- Classified as operating-system independent
- Test structure
- Repository includes pytest suites for the event bus, handler timeouts, and semaphores
Read from README.md, pyproject.toml, bubus/models.py, bubus/helpers.py, bubus/logging.py, bubus/service.py, bubus/__init__.py, tests/conftest.py, tests/test_eventbus.py, tests/test_semaphores.py, tests/test_handler_timeout.py.
What it can do
Process events asynchronously
Event messages and async handlers → Processed events with async execution
Process events synchronously
Event messages and sync handlers → Processed events with sync execution
Handle events in FIFO order
Sequential event queue → Events processed in first-in-first-out order
Handle events with concurrent processing
Multiple event messages → Events processed simultaneously across multiple threads/workers
Persist events using Write-Ahead Log
Event data and messages → Durable event storage with WAL persistence
Route events to registered handlers
Event messages and handler registrations → Events delivered to appropriate handler functions
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.