HTTPX API Rules

x-aid: httpx:httpx-rules
x-name: HTTPX Rules
description: >-
  Rules and guidelines for using the HTTPX Python HTTP client library
  effectively in synchronous and asynchronous applications.
x-type: Rules
x-tags:
  - Async
  - HTTP Client
  - Python
  - Rules
rules:
  - name: Reuse Clients
    description: >-
      Instantiate a single httpx.Client or httpx.AsyncClient and reuse it
      across requests. Creating a new client per request wastes connection
      pooling and TLS resources.
  - name: Use Async Client For Concurrency
    description: >-
      Use httpx.AsyncClient inside async code paths. Mixing sync and async
      transports inside the same event loop can block the loop and degrade
      performance.
  - name: Set Explicit Timeouts
    description: >-
      Always configure connect, read, write, and pool timeouts. HTTPX
      enforces strict timeouts and the default may not match your service
      level requirements.
  - name: Verify TLS By Default
    description: >-
      Leave certificate verification enabled. Disable it only for trusted
      development scenarios and never in production traffic.
  - name: Stream Large Responses
    description: >-
      Use the stream interface for large response bodies to avoid loading
      the entire payload into memory.
  - name: Handle Redirects Explicitly
    description: >-
      Redirects are not followed by default. Enable follow_redirects when
      you want requests behavior, and bound max_redirects to avoid loops.
  - name: Use HTTP/2 When Beneficial
    description: >-
      Enable http2=True on the client when talking to HTTP/2 capable
      origins for multiplexed performance gains. Install the http2 extra.
  - name: Pin The Library Version
    description: >-
      Pin httpx in requirements files. The library is pre 1.0 in some
      tracks and minor releases can introduce breaking changes.
x-maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com