Skip to main content

fused.options

fused.options​

List global configuration options.

This object contains a set of configuration options that control global behavior of the library. This object can be used to modify the options.

Examples:

Change the request_timeout option from its default value to 60 seconds:

fused.options.request_timeout = 60

Options​

dir​

__dir__() -> List[str]

base_url​

base_url: str = PROD_DEFAULT_BASE_URL

Fused API endpoint

shared_udf_base_url​

shared_udf_base_url: str = PROD_SHARED_UDF_DEFAULT_BASE_URL

Shared UDF API endpoint

auth​

auth: AuthOptions = Field(default_factory=AuthOptions)

Options for authentication.

show​

show: ShowOptions = Field(default_factory=ShowOptions)

Options for object reprs and how data are shown for debugging.

max_workers​

max_workers: int = 16

Maximum number of threads, when multithreading requests

run_timeout​

run_timeout: float = 130

Request timeout for UDF run requests to the Fused service

request_timeout​

request_timeout: Union[Tuple[float, float], float, None] = 5

Request timeout for the Fused service

May be set to a tuple of connection timeout and read timeout

metadata_request_timeout​

metadata_request_timeout: float = 60.0

Request timeout for file metadata requests (e.g., /file-metadata endpoint). These requests may involve processing large parquet files and can take longer.

request_max_retries​

request_max_retries: int = 5

Maximum number of retries for API requests

request_retry_base_delay​

request_retry_base_delay: float = 1.0

Base delay before retrying a API request in seconds

realtime_client_id​

realtime_client_id: Optional[StrictStr] = None

Client ID for realtime service.

max_recursion_factor​

max_recursion_factor: int = 5

Maximum recursion factor for UDFs. This is used to limit the number of recursive calls to UDFs. If a UDF exceeds this limit, an error will be raised.

save_user_settings​

save_user_settings: StrictBool = True

Save per-user settings such as credentials and environment IDs.

default_udf_run_engine​

default_udf_run_engine: Optional[StrictStr] = None

Default engine to run UDFs, one of: "local" or "remote".

default_validate_imports​

default_validate_imports: StrictBool = False

Default for whether to validate imports in UDFs before run_local, run_batch.

prompt_to_login​

prompt_to_login: StrictBool = False

Automatically prompt the user to login when importing Fused.

no_login​

no_login: StrictBool = False

If set, Fused will not attempt to login automatically when needed.

pyodide_async_requests​

pyodide_async_requests: StrictBool = False

If set, Fused is being called inside Pyodide and should use pyodide for async HTTP requests.

cache_directory​

cache_directory: Path | None = None

The base directory for storing cached results.

data_directory​

data_directory: Path = None

The base directory for storing data results. Note: if storage type is 'object', then this path is relative to fd_prefix.

temp_directory​

temp_directory: Path = Path(tempfile.gettempdir())

The base directory for storing temporary files.

never_import​

never_import: StrictBool = False

Never import UDF code when loading UDFs.

gcs_secret​

gcs_secret: str = 'gcs_fused'

Secret name for GCS credentials.

gcs_filename​

gcs_filename: str = '/tmp/.gcs.fused'

Filename for saving temporary GCS credentials to locally or in rt2 instance

gcp_project_name​

gcp_project_name: Optional[StrictStr] = None

Project name for GCS to use for GCS operations.

logging​

logging: StrictBool = Field(default=False, validate_default=True)

Control logging for Fused

verbose_udf_runs​

verbose_udf_runs: StrictBool = True

Whether to print logs from UDF runs by default

default_run_headers​

default_run_headers: Optional[Dict[str, str]] = {
    "X-Fused-Cache-Disable": "true"
}

(Advanced) Default headers to include with UDF run requests.

default_dtype_out_vector​

default_dtype_out_vector: StrictStr = 'parquet'

Default transfer type for vector (tabular) data

default_dtype_out_raster​

default_dtype_out_raster: StrictStr = 'npy,tiff'

Default transfer type for raster data

default_dtype_out_simple​

default_dtype_out_simple: StrictStr = 'json'

Default transfer type for simple Python data (bool, int, float, str, list)

fd_prefix​

fd_prefix: Optional[str] = None

If set, where fd:// scheme URLs will resolve to. By default will infer this from your user account.

verbose_cached_functions​

verbose_cached_functions: StrictBool = True

Whether to print logs from cache decorated functions by default

local_engine_cache​

local_engine_cache: StrictBool = True

Enable UDF cache with local engine

default_send_status_email​

default_send_status_email: StrictBool = True

Whether to send a status email to the user when a job is complete.

cache_storage​

cache_storage: StorageStr = 'auto'

Specify the default cache storage type

use_process_pool_for_local_submit​

use_process_pool_for_local_submit: StrictBool = False

Use ProcessPoolExecutor instead of ThreadPoolExecutor for local engine submit. This avoids GDAL thread-safety issues and Python GIL limitations, but has higher memory overhead and slower startup time.

row_group_batch_size​

row_group_batch_size: int = 32768

Target size in bytes for combining adjacent row group downloads. Adjacent row groups from the same file will be combined into single downloads until this threshold is reached. Default is 32KB (32768 bytes), which is optimized for S3 performance.

base_web_url​

base_web_url

save​

save()

Save Fused options to ~/.fused/settings.toml. They will be automatically reloaded the next time fused-py is imported.