## Summary
New `enable_logging=True` parameter on `responder.API()` that provides
structured, request-scoped logging using stdlib `logging` and
`contextvars`.
### What it does
- **`api.log`** — always available on every API instance. Works as a
plain logger by default; gains per-request context enrichment with
`enable_logging=True`
- **Per-request context** — every log message automatically includes
request ID, HTTP method, path, and client IP
- **Access logging** — logs every request with timing: `GET /path → 200
(1.2ms)`
- **Request ID** — generates or forwards `X-Request-ID` headers
(supersedes `request_id=True` when both are set)
- **stdlib logging** — works with any existing handler, formatter, or
log aggregator
### Usage
```python
# api.log always works — no setup required
api = responder.API()
api.log.info("starting up") # plain logger, no context
# With enable_logging=True, log messages get request context automatically
api = responder.API(enable_logging=True)
@api.route("/")
def index(req, resp):
api.log.info("handling request")
# => 2026-03-24 12:00:00 [INFO] responder.app — handling request [GET /] [req:abc123] [client:127.0.0.1]
```
For additional loggers in helper modules:
```python
from responder.ext.logging import get_logger
logger = get_logger("myapp.db")
```
### Architecture
- `responder/ext/logging.py` — self-contained module with:
- `LoggingMiddleware` — pure ASGI middleware that sets contextvars and
logs access
- `RequestContextFilter` — logging filter that injects context into
records
- `RequestContext` — read-only access to current request metadata
- `get_logger()` / `setup_logging()` — convenience functions
- `api.log` — always a valid logger; context-aware when
`enable_logging=True`, plain stdlib logger otherwise
- Wired into `API.__init__` via the `enable_logging` parameter
### Files
- `responder/ext/logging.py` — new module
- `responder/api.py` — added `enable_logging` parameter and `api.log`
- `tests/test_logging.py` — 9 tests
- `docs/source/tour.rst` — new Structured Logging section
- `docs/source/index.rst` — added to feature list
## Test plan
- [x] 9 logging tests pass
- [x] Full suite: 208 passed
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- **Deployment guide**: health check endpoint, Docker Compose example,
Caddy reverse proxy config, Procfile pattern, production checklist
- **API reference**: quick usage examples for every class (API, Request,
Response, RouteGroup, BackgroundQueue, RateLimiter, status helpers)
- **Feature tour**: new Pydantic validation and content negotiation
sections, expanded MessagePack docs
- **Testing guide**: rate limiting and mounted WSGI app test examples,
Werkzeug 3.1.7 tip
- **Middleware tutorial**: pure ASGI middleware example (no
BaseHTTPMiddleware dependency)
- **CLI guide**: environment variables section (PORT, SECRET_KEY)
- **Homepage**: updated feature list with SSE, rate limiting, Pydantic,
content negotiation, route groups
- **Backlog**: removed already-implemented items, added current ideas
+362 lines of docs, no code changes.
## Test plan
- [x] `make html` builds cleanly with no warnings
- [x] All 199 tests pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Signed-off-by: Kenneth Reitz <me@kennethreitz.org>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Andreas Motl <andreas.motl@panodata.org>
- Add docstrings to all undocumented public methods across API, Request,
Response, Router, Route, BackgroundQueue, and related classes
- Expand api.rst with autodoc sections for RouteGroup, BackgroundQueue,
QueryDict, and RateLimiter
- Update starlette dependency to >=1.0
- Drop Python 3.9 support (required by Starlette 1.0), minimum is now 3.10
- Bump version to 3.4.0
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New documentation pages:
- Authentication: API keys, JWT tokens, session auth, custom exceptions
- WebSocket Tutorial: echo server, chat room, HTML client, data formats
- Writing Middleware: hooks vs middleware, Starlette integration, ordering
- Configuration: env vars, .env files, secret keys, debug mode, production setup
Also:
- Complete working example at end of quickstart with cross-references
- Three new example files: rest_api.py, websocket_chat.py, sse_stream.py
- Changelog updated for v3.2.0
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Three new tutorial pages:
- Building a REST API: full CRUD with Pydantic validation, from scratch
- Using SQLAlchemy: async engine, lifespan setup, CRUD with ORM
- Migrating from Flask: concept mapping, quick reference table,
gradual migration via app mounting
Also rewritten:
- CLI docs: cleaner, more concise
- API reference: added prose descriptions for each section
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Define your API schemas with Pydantic models instead of (or alongside)
YAML docstrings and marshmallow:
from pydantic import BaseModel
class PetIn(BaseModel):
name: str
age: int = 0
class PetOut(BaseModel):
id: int
name: str
age: int
@api.route("/pets", methods=["POST"],
request_model=PetIn, response_model=PetOut)
async def create_pet(req, resp):
data = await req.media()
resp.media = {"id": 1, **data}
Also works with @api.schema("Name") decorator for registering
standalone schema components.
Pydantic models, marshmallow schemas, and YAML docstrings can all
be used together in the same API.
Also: rewrite docs with more prose, restore sidebar logo and links,
add FastAPI acknowledgment, update homepage copy.
161 tests, 95% coverage.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
A few of the adjustments here have been required to mitigate Sphinx
warnings, which would converge to errors on CI, thus failing the build.
A few other changes, both wording and syntax/formatting fixes, are
coming from regular copyediting and documentation maintenance.
Basically linked things I had to google/look up on my own. There are a million things that _could_ be linked, but using myself as a median Python coder, it seems about right. Removed italics because I can't figure out how to use them with a link.
kennethreitz
Andreas Motl