v1.2.0

• Added wait_max argument to retryhttp.wait_from_header and retryhttp.wait_retry_after, which defaults to 120.0 seconds.
• retryhttp._utils.is_rate_limited: revert behavior to correctly determine rate limiting by a 429 Too Many Requests` status.
• When receiving 503 Service Unavailable, honor a Retry-After header if provided.
• Rename retryhttp.wait_rate_limited to retryhttp.wait_retry_after, but retain alias for backwards compatibility and convenience.
• retryhttp.wait_from_header: Handle case if server responds with a date in the past.
• retryhttp.wait_context_aware: The wait_server_errors argument now defaults to retryhttp.wait_retry_after with tenacity.wait_random_exponential as fallback, since some server errors may respond with a Retry-After` header.
• retryhttp.wait_context_aware: The wait_rate_limited argument now has tenacity.wait_random_exponential as fallback to retryhttp.wait_retry_after, to make retrying rate-limited requests more robust.

v1.1.0

• Add HTTP-date value parsing for retryhttp.wait_from_header
• retryhttp._utils.is_rate_limited now determines that a request was rate limited by the presence of a Retry-After header. In prior versions, this was based on the status code 429 Too Many Requests. However, many servers return other status codes when rate limiting.

v1.0.1

• FIx documentation errors.

v1.0.0

• API is now stable. Any breaking changes will follow SemVer guidelines.
• Added requests.exceptions.ChunkedEncodingError to the list of default network errors.

v0.2.0

• Rename retryhttp.retry_http_errors to retryhttp.retry.
• Rename retryhttp.wait_http_errors to retryhttp.wait_context_aware.
• Restructure project so all members are part of the root namespace.
• Move type delcarations to retryhttp._types.
• Enable bare decorator syntax for retryhttp.retry (i.e., @retry works as well as @retry())