_constants.py

# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.

import httpx

RAW_RESPONSE_HEADER = "X-Stainless-Raw-Response"
OVERRIDE_CAST_TO_HEADER = "____stainless_override_cast_to"

# default timeout is 30 seconds
DEFAULT_TIMEOUT = httpx.Timeout(timeout=30, connect=5.0)
DEFAULT_MAX_RETRIES = 5
DEFAULT_CONNECTION_LIMITS = httpx.Limits(max_connections=20, max_keepalive_connections=10)

# Separate H2 connection pools by workload. Each "shard" is its own httpx
# client / transport (≈ one H2 connection). Defaults sized for Jetty's ~128
# streams/connection: enough API/wait concurrency without oversized transfer.
DEFAULT_API_POOL_SHARDS = 8
DEFAULT_BACKGROUND_POOL_SHARDS = 16
DEFAULT_TRANSFER_POOL_SHARDS = 2

INITIAL_RETRY_DELAY = 1.0
MAX_RETRY_DELAY = 60.0

# Long-poll endpoints tell the server how long to hold the connection and expect a
# graceful 408 when that hold elapses. The per-request client read timeout must
# exceed the server hold so the server returns 408 first instead of the client
# aborting the stream (RST_STREAM) and surfacing a connection error.
LONG_POLL_CLIENT_BUFFER_SECONDS = 5.0

# Authoritative server-side long-poll hold clamps, mirrored from the mux controllers.
STATUS_LONG_POLL_SERVER_MAX_SECONDS = 30.0
EXEC_LONG_POLL_SERVER_MAX_SECONDS = 25.0

# Maximum allowed size (in bytes) for individual entries in `file_mounts` when creating Blueprints
# NOTE: Empirically, ~131,000 is the maximum command length after
# base64 encoding; 98,250 is the pre-encoded limit that stays within that bound.
# We measure size in bytes using UTF-8 encoding; base64 output is ASCII.
FILE_MOUNT_MAX_SIZE_BYTES = 98_250

# Maximum allowed total size (in bytes) across all `file_mounts` when creating Blueprints
FILE_MOUNT_TOTAL_MAX_SIZE_BYTES = 786_000 * 10  # ~10 mb