NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI, #275 most downloaded on PyPI
Snowflake DB driver for Python
Latest stable
5.2.0
Last release 18 days ago
04 Nov 2026
Ships fairly regularly
a new release about every 2 weeks
Nearly every release is documented
notes for 59 of the last 60 stable releases
6 versions withdrawn
withdrawn after publishing
9 years old
193 releases · first in 2017
Drag across the ruler to compare, or tap a release to open it.
Added support for Python 3.15 and free-threaded Python 3.15t. The pandas extra installs pandas 3 on Python 3.15. (snowflakedb/drivers#2929)
New features:
force_microsecond_precision setting so Arrow and pandas fetches use it when the call omits the argument. (snowflakedb/drivers#3052)EllipticCurvePrivateKey objects as private_key. (snowflakedb/drivers#3005)~ in token_file_path, so a path such as ~/.snowflake/pat.token is read from the home directory when loading a PAT, OAuth, or OIDC bearer token. (snowflakedb/drivers#3049)Changes:
[pandas] extra to allow installation of pandas 3.0.0 and later. (snowflakedb/drivers#2971)split_statements to apply one delimiter for the whole script, read when the call starts. (snowflakedb/drivers#3043)Bug fixes:
cursor.execute() raising 000006: Invalid Snowflake response for an empty or whitespace-only statement; it now skips the statement with a warning and returns the cursor, as the 4.x connector did. (snowflakedb/drivers#2970)Error.errorhandler_make_exception so it constructs only PEP 249 Error subclasses. (snowflakedb/drivers#2873)cursor(DictCursor) remains a DictCursor through execute and executemany, its fetch results and next() are dictionaries, and the default cursor() is narrowed to a tuple cursor, for both the sync and async cursors. (snowflakedb/drivers#2872)service_name connection property, reading the live SERVICE_NAME session parameter from the core driver so the value tracks cluster reassignments made by query responses. (snowflakedb/drivers#3068)write_pandas accepted column names that are not valid unquoted identifiers when quote_identifiers=False. (snowflakedb/drivers#2977)validate_default_parameters=true not failing login when the specified database, schema, or warehouse does not exist. (snowflakedb/drivers#2974)Your coding agent can read these notes before it upgrades. Set up the MCP server →
Added support for Python 3.14t (free-threaded CPython).
New features:
Bug fixes:
Changes:
CLIENT_SESSION_KEEP_ALIVE and CLIENT_PREFETCH_THREADS unless the caller sets them.CLIENT_SESSION_KEEP_ALIVE, CLIENT_PREFETCH_THREADS, QUERY_TAG, and a heartbeat frequency clamped to 900–3600 seconds as session parameters, and to send a set validate_default_parameters value, true or false, as CLIENT_VALIDATE_DEFAULT_PARAMETERS, leaving that parameter off the login request when it is unset.TIMEZONE, AUTOCOMMIT, CLIENT_TIMESTAMP_TYPE_MAPPING, CLIENT_METADATA_REQUEST_USE_CONNECTION_CTX, SERVICE_NAME, GCS_USE_DOWNSCOPED_CREDENTIAL, and CLIENT_OUT_OF_BAND_TELEMETRY_ENABLED as session parameters when they are set.disable_console_login and oauth_disable_console_login. Passing either connection parameter warns and has no effect, and external browser login always requests the SSO URL from Snowflake.OCSP_MODE from the login client environment and from connection diagnostics. The driver does not perform OCSP; certificate revocation continues to be reported as CERT_REVOCATION_CHECK_MODE.Content-Encoding header.…a process pool, and client_fetch_threads is a deprecated alias of client_prefetch_threads .
Breaking changes:
connections.toml so connections defined in config.toml are kept.ConfigManager with no file_path so reading an option raises MissingConfigOptionError instead of ConfigManagerError.connect() so a detected application name — from the SF_PARTNER environment variable, or from an imported Streamlit, Jupyter, or snowbooks module — takes precedence over application in the default connection profile.client_session_keep_alive and client_session_keep_alive_heartbeat_frequency to read-only properties; assigning them raises AttributeError.snowflake.connector.config_manager.CONFIG_PARSER; use CONFIG_MANAGER.ConfigManager.add_subparser() and ConfigManager._sub_parsers; use add_submanager() and _sub_managers.Connection.connect() method; the session is opened when the connection is constructed.region connection parameter and Connection.region; pass a fully qualified account or set host instead.backoff_policy callable; retry delays are set with retry_backoff_base_ms, retry_backoff_cap_ms, retry_backoff_factor, and retry_backoff_jitter, the default initial delay is 250 ms (was 1 s), and the default jitter is decorrelated (was full).Connection.disable_request_pooling; passing disable_request_pooling to connect() warns and has no effect.converter_class connection parameter and DefaultConverterClass; passing converter_class raises ProgrammingError.invalid_client, in the exception message.master_token is required and invalid tokens fail in connect() instead of on a later request.http:// is allowed only for loopback addresses.ReauthenticationRequest instead of opening the browser again and retrying the query.disable_console_login warns and has no effect: the parameter defaults to false, and the driver always requests the SSO URL from Snowflake instead of opening /console/login.password_callback connection parameter; passing it raises ProgrammingError, and an expired password must be changed outside the driver.InterfaceError; fetchone, fetchmany, fetchall, and reset still read rows already buffered on an open cursor after connection.close().fetchone, fetchmany, fetchall, and cursor iteration before any query to raise ProgrammingError instead of TypeError.cursor.execute so passing file_stream on a statement that is not a PUT raises ProgrammingError.fetch_pandas_all, fetch_pandas_batches, and the fetch_arrow_* methods so a JSON result, such as SHOW or DESCRIBE, returns a DataFrame or Arrow table instead of raising NotSupportedError.executemany() on an INSERT without a VALUES clause to raise ProgrammingError instead of InterfaceError.create_temp_table argument from write_pandas; pass table_type='temp' instead.ResultBatch.session_manager and ResultBatch.http_config.client_fetch_use_mp warns and no longer selects a process pool, and client_fetch_threads is a deprecated alias of client_prefetch_threads... can select a different file.iobound_tpe_limit; passing it to connect() warns and has no effect, and Connection.iobound_tpe_limit does not exist.proxy_host/proxy_port or proxy URL wins when HTTP_PROXY, HTTPS_PROXY, or NO_PROXY is also set.enable_crl_cache=False warns and no longer disables it; set both enable_crl_memory_cache and enable_crl_file_cache to False to turn caching off.SNOWFLAKE_FIPS=1; see the installation instructions for details.SNOWFLAKE_TROUBLESHOOTING_ENABLED environment variable.snowflake.connector.options module; importing it raises ModuleNotFoundError.snowflake.connector.dbapi and the authenticator classes previously exported from snowflake.connector.auth.snowflake.connector.constants, including FileTransferType, ResultStatus, cloud filesystem labels, part-size constants, the get_*_types and is_*_type_name helpers, and the Arrow type helpers; QueryStatus remains.DownloadMetrics, RemoteChunkInfo, and the chunk-download constants DOWNLOAD_TIMEOUT, MAX_DOWNLOAD_RETRY, SSE_C_AES, SSE_C_ALGORITHM, and SSE_C_KEY from snowflake.connector.result_batch.New features:
Added an asyncio-compatible API under snowflake.connector.aio, including nonblocking connection, query, fetch, transaction, Arrow, pandas, and async context-manager operations.
Added the use_core_arrow connection parameter to opt into native Rust conversion for row, Arrow, and pandas result fetching in the synchronous and asyncio APIs.
Bug fixes:
Fixed PUT SOURCE_COMPRESSION=AUTO_DETECT so a .deflate file is recognized as DEFLATE.
Fixed GET so a failed download removes the temporary .part file.
Fixed Azure client-side encryption metadata so EncryptionAgent.EncryptionAlgorithm is AES_CBC_256.
Fixed the GET result file column so a stage path with more than one directory level reports only the file name.
Fixed Connection.is_valid() so a live PAT_WITH_EXTERNAL_SESSION connection returns true.
Changes:
SOURCE_COMPRESSION=BROTLI, including auto-detection of .br files.private_key_password; private_key_file_pwd is still accepted and warns.client_store_temporary_credential; client_request_mfa_token is still accepted and warns.connection.rest.request() to raise ValueError for an HTTP method other than get or post.JSONResultBatch and ArrowResultBatch so to_arrow(), to_pandas(), and table iteration work on every result batch.cn-*) to use the AWS SDK's default endpoint when no regional URL and no custom endpoint are set.oauth_enable_refresh_tokens so it warns and has no effect; a refresh token is used when the identity provider returns one, and client_store_temporary_credential controls whether that token is cached on disk.oauth_credentials_in_body=True so the client id and secret are sent only in the token-request body.use_s3_regional_url; enable_stage_s3_privatelink_for_us_east_1 is still accepted and warns.Content-Encoding and x-ms-blob-content-encoding headers.disable_parallel_user_prompt and client_store_temporary_credential are left at their defaults.executemany to follow the server's array-binding flag instead of detecting INSERT in the SQL text; a statement the server does not mark as array-bindable still runs one row at a time.snowflake.connector.telemetry_oob module.KeyWrappingMetadata.EncryptionLibrary is Rust(OpenSSL).parameter and validation_code on errors from connection-configuration validation.fetch_pandas_all and fetch_pandas_batches so extra keyword arguments are passed to Arrow's to_pandas; arguments that only pandas.concat accepts, such as sort, keys, and copy, raise TypeError.login_timeout, network_timeout, and socket_timeout so login_timeout returns the login budget (120 seconds when unset, previously None), network_timeout returns request_timeout (120 seconds when unset, previously None), and socket_timeout returns retry_timeout; passing network_timeout warns and also sets query_timeout, passing socket_timeout warns and also sets connect_timeout, and retry_timeout is a total time limit for the attempt rather than a per-read inactivity timeout.cursor._request_id to warn on use; it returns the same uuid.UUID as cursor.request_id.ProgrammingError with errno 255001.display_size for text columns, set to the column's character length.put_get_max_attempts so any positive integer is used, with no upper limit.client_store_temporary_credential to true on every platform; an explicit false now disables token caching on macOS and Windows as well.dsn, session_id, support_negative_year, json_result_force_utf8_decoding, inject_client_pause, probe_connection, debug_arrow_chunk, snowflake_server_dop_cap_for_file_transfer, and reraise_error_in_file_transfer_work_function so they warn and have no effect; connection.session_id after login still returns the server-assigned id, and log_imported_packages and log_imported_packages_in_telemetry are ignored without a warning.snowflake.connector.errors, warn on first use, and are no longer raised: HttpError, InternalServerError, ServiceUnavailableError, GatewayTimeoutError, ForbiddenError, RequestTimeoutError, BadRequest, BadGatewayError, MethodNotAllowed, TooManyRequests, OtherHTTPRetryableError, RevocationCheckError, RefreshTokenError, BindUploadError, RequestExceedMaxRetryError, TokenExpiredError, and PresignedUrlExpiredError.Added faster data conversion method, enabled via connection parameter use_core_arrow=True
use_core_arrow=TrueAdded Duo push continuation and MFA callbacks for username/password authentication.
DictCursor rows.SHOW, DESCRIBE, DDL, PUT, and GET.platform_detection_timeout_seconds. Setting it to 0 skips cloudput_compress_level and put_tempdir connection parameters for PUTproxy_scheme for HTTPS connections to proxies andextra_root_store_path for extending the default TLS trust store.executemany().FILE, INTERVAL_YEAR_MONTH, and INTERVAL_DAY_TIME type codes.client_store_temporary_credential to default to True. Set it toFalse explicitly to disable ID, OAuth, and MFA token caching.timezone, interpolate_empty_sequences, and reuse_resultscursor.description reporting GEOGRAPHY and GEOMETRY columns asALTER SESSION, COMMIT, and other statements rejected by statementALTER SESSION SET cache updates with non-ASCII SQL text.SNOWFLAKE_CONNECTIONS parsing so Snowpark configuration is read asReauthenticationRequest so it is importable fromsnowflake.connector.network and exposes its underlying cause.write_pandas() quoting for Parquet and Iceberg field and optionprivate_key.file column reports theSNOW-4075338: numpy scalar binding coverage
SNOW-4075338: numpy scalar binding coverage
## Summary
- Ports the legacy connector `test_numpy_datatype_binding` matrix into
`python/tests/e2e/query/test_numpy_binding.py` (numpy integers, floats,
bools, and `datetime64` into NTZ/DATE/LTZ/TZ).
- Adds unit converter cases for those numpy scalar widths and the legacy
`datetime64` values.
- Adds a pyformat e2e case for the SNOW-4075338 nanosecond
`numpy.datetime64` bind.
These tests currently fail on `datetime64` with `ProgrammingError
255001` (the 5.x regression). Numeric and bool numpy binds already pass.
No production fix in this PR.
## Test plan
- [ ] Confirm unit `TestClientSideBindingConverterNumpy` /
`TestJsonBindingConverterNumpy` run locally.
- [ ] Confirm e2e `test_numpy_binding.py` and
`test_should_bind_numpy_datetime64_with_positional_pyformat` fail only
on datetime64 until the converter is fixed.
- [ ] After the converter fix, re-run the same tests and expect them to
pass.
Made with [Cursor](https://cursor.com)
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
GitOrigin-RevId: f5b7666
JDBC Driver 5.x is in public preview and available to all accounts. Validate it against a non-production account. Preview builds are not covered by th
JDBC Driver 5.x is in public preview and available to all accounts. Validate it against a non-production account. Preview builds are not covered by the client support policy in Client versions & support policy. Use of this preview is subject to Snowflake’s Preview Terms.
See the changelog
JDBC Driver 5.x is in public preview and available to all accounts. Validate it against a non-production account. Preview builds are not covered by th
JDBC Driver 5.x is in public preview and available to all accounts. Validate it against a non-production account. Preview builds are not covered by the client support policy in Client versions & support policy. Use of this preview is subject to Snowflake’s Preview Terms.
snowflake.connector.aio for non-blocking connect,
execute, fetch, and concurrent queries on a single thread.ConnectionConfig, generated from the shared parameter registry,
passable via connect(config=...) or built automatically from kwargs.sf_core) used across universal-driver
bindings.InterfaceError (errno=252006) on all operations;
previously only execute() was guarded and other methods failed with
TypeError / AttributeError.ProgrammingError
(errno=252009) instead of TypeError.Connection.connect() method; the session is established
at construction time—create a new connection to reconnect.region connection parameter and Connection.region; use a
fully qualified account or set host explicitly.client_fetch_threads (alias of client_prefetch_threads);
client_fetch_use_mp is accepted but has no effect (always a thread pool).client_session_keep_alive and
client_session_keep_alive_heartbeat_frequency read-only; assignment raises
AttributeError—set them at connection time.oauth_enable_refresh_tokens (deprecated no-op); refresh tokens are
always used when the IdP returns one—gate disk caching with
client_store_temporary_credential.enable_stage_s3_privatelink_for_us_east_1 to use_s3_regional_url
(old name remains a deprecated alias).HTTP_PROXY / HTTPS_PROXY / NO_PROXY by default; set
use_proxy_env=True (or PROXYWITHENV=true) to opt in. Explicit proxy
params win over env vars.ReauthenticationRequest on master-token expiry (Snowpark
reauth signal not yet wired).amazonaws.com.cn S3 endpoint for cn-*
regions; the AWS SDK resolver is used unless an explicit or regional-URL
override is set.Content-Encoding: utf-8 on Azure Blob PUT (matches JDBC /
libsnowflakeclient).Nothing published for this version
Nothing published for this version
⚠️ PRIVATE PREVIEW DISCLAIMER ⚠️ This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snowflake's Previe
⚠️ PRIVATE PREVIEW DISCLAIMER ⚠️ This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snowflake's Preview Terms. This driver is not ready for use with production data and may be unstable. Please only use it if you are actively participating in the Universal Driver Private Preview.
⚠️ PRIVATE PREVIEW DISCLAIMER ⚠️ This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snowflake's Previe
⚠️ PRIVATE PREVIEW DISCLAIMER ⚠️ This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snowflake's Preview Terms. This driver is not ready for use with production data and may be unstable. Please only use it if you are actively participating in the Universal Driver Private Preview.
:warning: PRIVATE PREVIEW DISCLAIMER :warning: This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snow
:warning: PRIVATE PREVIEW DISCLAIMER :warning: This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snowflake's Preview Terms. This driver is not ready for use with production data and may be unstable. Please only use it if you are actively participating in the Universal Driver Private Preview.
:warning: PRIVATE PREVIEW DISCLAIMER :warning: This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snow
:warning: PRIVATE PREVIEW DISCLAIMER :warning: This is a beta version of our Universal Driver intended for Snowflake customers who are subject to Snowflake's Preview Terms. This driver is not ready for use with production data and may be unstable. Please only use it if you are actively participating in the Universal Driver Private Preview.
Nothing published for this version
OCSP certificate revocation checks are now off unless you opt in. Set ocsp_fail_open=True or ocsp_fail_open=False to enable OCSP. The stored default i
OCSP certificate revocation checks are now off unless you opt in. Set ocsp_fail_open=True or ocsp_fail_open=False to enable OCSP. The stored default is None (unset); only an explicit True/False opts in, so forwarding DEFAULT_CONFIGURATION as kwargs does not turn OCSP on. disable_ocsp_checks=True (or insecure_mode=True) always turns OCSP off, including when ocsp_fail_open is also set. disable_ocsp_checks=False and insecure_mode=False are the stored defaults and are not an opt-in. Connection attribute ocsp_fail_open is no longer a report of whether OCSP is fail-open; it is the stored preference (None = unset, True = fail-open, False = fail-closed). Use _ocsp_mode() / disable_ocsp_checks to see whether checks are actually on. ocsp_response_cache_filename and ocsp_root_certs_dict_lock_timeout do not turn OCSP on; if they are set without an OCSP mode parameter they are ignored and a warning is logged. The SF_OCSP_FAIL_OPEN environment variable still only switches fail-open vs fail-closed after OCSP is already on. Login OCSP_MODE telemetry now reports DISABLE_OCSP_CHECKS by default. The process-global FEATURE_OCSP_MODE is updated by each constructed REST client, except that a later default (OCSP off) client does not overwrite a non-default already stored on the process, and a later FAIL_OPEN client does not overwrite FAIL_CLOSED.
Fixed external-browser (SSO) authentication to validate the Origin header on the local callback server, rejecting tokens delivered from unexpected origins. A trailing slash in the origin (e.g. https://account.snowflakecomputing.com/) is now accepted on par with the bare origin, matching JDBC and other driver behaviour. Preconnect probe connections (empty recv) no longer count against the retry budget and no longer abort the login flow.
Added the SNOWFLAKE_TLS_CIPHERS environment variable to restrict which TLS ciphers the connector offers. It takes a colon-separated list; names beginning with TLS_ are applied as TLS 1.3 cipher suites and the remainder as the cipher list for TLS 1.2 and below, so a single variable covers both. Leaving it unset keeps OpenSSL's defaults unchanged, and an unrecognized cipher name is rejected rather than silently ignored. The restriction covers Snowflake API traffic, cloud-storage (stage) transfers, OCSP/CRL fetches and IdP requests. Requests issued by the AWS and Azure SDKs, and asynchronous connections, are not covered — for TLS 1.3 suites specifically they cannot be, because the Python standard library exposes no API for restricting them.
Raised the minimum pyOpenSSL requirement to 25.3.0, the first version providing set_tls13_ciphersuites. This does not narrow the set of installable versions in practice: earlier releases cap cryptography below 46 and so were already uninstallable alongside the connector's own cryptography>=46.0.5 requirement.
Added the workload_identity_host connection option that overrides the STS host used by AWS Workload Identity Federation, for endpoints the driver cannot derive from the region (such as an interface VPC endpoint). The default STS host is now resolved via botocore so partitions that do not use amazonaws.com (ISO, European Sovereign Cloud, ...) get the correct hostname. A privately routed host cannot be reached by Snowflake on the default GetCallerIdentity path, so a VPC or PrivateLink STS endpoint also requires workload_identity_aws_use_outbound_token=True (SNOW-4017192).
Fixed MD5 computation for Azure clouds (SNOW-4168830).
Follow-up to the v4.7.4 incomplete-result fix ( SNOW-4109042 ): when a successful query-request has an incomplete inline first chunk, the connector re
connect() as a generic 250001: Could not connect to Snowflake backend after N attempt(s) alongside a firewall-troubleshooting hint. The network layer already identified and named such failures, but the authentication layer retried them until the login timeout expired and then replaced the diagnosis. They now surface as NonRetryableTlsError (a subclass of OperationalError keeping the same errno, so existing handlers are unaffected) naming the underlying cause, and are no longer retried. Transient handshake faults (ECONNRESET, unexpected EOF) remain retryable.SNOWFLAKE_MIN_TLS_VERSION environment variable (1.2, 1.3, TLSv1.2 or TLSv1.3, defaulting to TLS 1.2) to raise the minimum TLS version on every outbound connection a synchronous connector makes: the Snowflake API, stage transfers, OCSP/CRL fetches, platform detection, IdP requests, AWS SDK requests (workload identity STS calls and the platform-detection identity probe) and Azure AD token requests made by azure-identity. An unrecognized value is rejected at connect time.Fixed missing retries on transient HTTP failures when fetching an OAuth access token from the IdP token endpoint ( OAUTH_CLIENT_CREDENTIALS and OAUTH_
OAUTH_CLIENT_CREDENTIALS and OAUTH_AUTHORIZATION_CODE). Token requests now retry transport errors, HTTP 408/429, and 5xx responses (SNOW-3984430, #3003).SSLError containing SysCallError(-1, 'Unexpected EOF')) was classified as non-retryable and surfaced as an OperationalError, unlike ECONNRESET. Such handshake Unexpected EOF errors are now retried, on both the sync and async request paths (SNOW-4058589).fetchone(), fetchmany(), fetchall() and cursor iteration silently returning an incomplete result set. A downloaded result chunk holding fewer rows than the back-end reported just ended the iteration, which the fetch methods report as a normal end of results, so callers received fewer rows than the query produced with no exception raised. Such a chunk now raises an OperationalError (errno 252013). Errors raised while iterating a result set are no longer reported as end-of-results either: _fetchone() caught every TypeError and returned None, so a failure anywhere in the download/parse chain was indistinguishable from an exhausted result set. Both the sync and async paths are fixed (SNOW-4109042).Added experimental Python 3.14t (free-threaded CPython) wheel support. Experimental — not intended for production use.
connect() being significantly slower on deep call stacks (e.g. Django apps with several middleware/decorator layers) because get_application_path() used inspect.stack(), which reads and parses the source file of every frame on the stack. It now walks frame references directly instead (SNOW-3691001, #2908).split_statements truncating unquoted URLs containing :// at the //, misreading it as a line comment. The guards added alongside // comment support in 4.7.2 (SNOW-3772985) only recognized file://, so every other scheme (http://, https://, s3://, snow://, azure://) was cut short. Any scheme:// is now recognized, on both the line-comment and block-comment paths (the latter affected globs such as s3://bucket/*.csv) (SNOW-3930192).RuntimeError: UnixFileLock on ... was inherited across fork; construct a new instance when it was written from a forked child process. This could surface on fork-based platforms when fetching results with client_fetch_use_mp=True, where chunk downloads run in worker processes that perform their own certificate revocation checks. The cache's file lock is now re-created when the owning process changes, matching what already happened for the spawn start method.qrmk is present, for both the sync and async result paths (SNOW-3675590).create_batches_from_response() reports chunk-header names with type/length value metadata (SNOW-3675590).create_batches_from_response() reports a structural summary of the response (SNOW-3675590)._format_query_for_log() runs the query through SecretDetector, covering the cursor and connection paths on both sync and async. The cancel-query log paths use the same masking (SNOW-3675590)._format_query_for_log_lazy() wrapper whose __str__ masks only whena handler emits the record (SNOW-3675590).WORKLOAD_IDENTITY authenticator runs only against recognized Snowflake hosts: the connection host is normalized and suffix-anchored against snowflakecomputing.com, .cn, and .mil before the workload-identity flow proceeds. The SNOWFLAKE_WIF_ALLOWED_HOST_SUFFIXES environment variable additively extends the recognized-host list (SNOW-3675580).SF_OCSP_RESPONSE_CACHE_SERVER_URL is set, the connector uses it as the OCSP cache server, including on PrivateLink hosts..privatelink. label, and end at snowflakecomputing.com, .cn, or .mil. Other hosts use the public default cache URL; set SF_OCSP_RESPONSE_CACHE_SERVER_URL to use a different cache server.: and % in the token value. Added patterns for OAuth access and refresh tokens in serialized JSON, for one-time passcodes, and for OAuth client identifiers and secrets. The passcode pattern accepts quoted values. The password pattern requires at least six characters in the value, so ordinary prose following the word "password" is not redacted.Authorization header in error logs (SNOW-3675593).snowflake.connector loggers by default, independent of how application logging is configured. The same filter also covers the third-party loggers the connector uses (botocore, boto3, aiohttp, aiobotocore, aioboto3, vendored urllib3). Set SNOWFLAKE_DISABLE_LOG_SECRET_MASKING=true to opt out; it is honored at emit time, including after import (SNOW-3675583).Fixed a thread leak in the file transfer agent by properly shutting down ThreadPoolExecutors after PUT/GET transfers ( SNOW-3556240 , #2878 ).
split_statements treating // as SQL instead of a line comment, which could merge multiple statements when a // comment contained an apostrophe (SNOW-3772985).250001 Invalid OAuth access token when the cached token was invalid (GS code 390303); the connector now reauthenticates silently (via refresh token if available, otherwise browser) instead of hard-failing. Also fixed fresh processes holding only a cached refresh token going straight to an interactive browser prompt instead of attempting a silent refresh first.login_timeout as 250003: Failed to execute request: Attempted to set connect timeout to <negative value>. Obtaining the one-time token in step 4 can overshoot the login deadline, which made the remaining budget negative; it was then passed to the HTTP layer, which rejected it with an opaque ValueError. A timed-out SAML login now raises 250006 with a message naming login_timeout. Running out of budget while retrying on RefreshTokenError reports the same error instead of failing later with an AttributeError on an empty response (SNOW-3891419).urlparse().hostname always returns lowercase, but the host was compared case-sensitively in _is_snowflake_as_idp, causing Snowflake-as-IdP detection to return False and raising error 251013 (client_id is empty) even for connections that don't require a client ID.Added support for Python 3.14t (free-threaded).
win_arm64 (no cryptography wheels available) and mitmproxy proxy tests on all platforms (transitive dependencies aioquic/pylsqpack lack free-threaded-compatible wheels).python-connector.log not rotating on Windows, and every record being logged twice, when easy logging is enabled via config.toml (SNOW-3680325).
logging.basicConfig() and therefore no longer configures the root logger. python-connector.log now captures only the snowflake.connector, botocore, and boto3.is_valid_url() with urllib.parse.urlparse (SNOW-3392651).reauthenticate() calls _request_tokens() directly instead of looping through prepare(). Token cache is now read exactly once per connection, and _store_tokens() preserves macOS Keychain ACL by never calling remove(). The async OAuth reauthenticate() now runs the synchronous OAuth flow on a worker thread instead of blocking the event loop.offline_access scope for token endpoints on *.snowflakecomputing.com or *.snowflakecomputing.cn, which caused invalid_scope errors. Snowflake custom OAuth expects refresh_token in scope instead. External IdP behavior is unchanged.scale metadata in Arrow result set processing for TIME, TIMESTAMP_NTZ, TIMESTAMP_LTZ, and TIMESTAMP_TZ columns (SNOW-3388299).x-amz-bucket-region response header on each redirect. Redirects are capped at 5 hops.WorkloadIdentityCredential to authenticate via the injected service account credentials. OIDC backward compatibility is also supported.workload_identity_aws_use_outbound_token connection option (default false) to opt into AWS WIF JWT attestation via STS GetWebIdentityToken instead of the default SigV4 GetCallerIdentity method.CREATE VIEW db.schema.obj) on a session with no current schema would populate the connector's cached _schema/_database from the referenced object's namespace. This made get_current_schema() diverge from the server's CURRENT_SCHEMA() and mis-qualified Snowpark temp objects (SNOW-3665226).Fixed python-connector.log not rotating on Windows, and every record being logged twice, when easy logging is enabled via config.toml (SNOW-3680325).
python-connector.log not rotating on Windows, and every record being logged twice, when easy logging is enabled via config.toml (SNOW-3680325).
logging.basicConfig() and therefore no longer configures the root logger. python-connector.log now captures only the snowflake.connector, botocore, and boto3.is_valid_url() with urllib.parse.urlparse (SNOW-3392651).reauthenticate() calls _request_tokens() directly instead of looping through prepare(). Token cache is now read exactly once per connection, and _store_tokens() preserves macOS Keychain ACL by never calling remove(). The async OAuth reauthenticate() now runs the synchronous OAuth flow on a worker thread instead of blocking the event loop.offline_access scope for token endpoints on *.snowflakecomputing.com or *.snowflakecomputing.cn, which caused invalid_scope errors. Snowflake custom OAuth expects refresh_token in scope instead. External IdP behavior is unchanged.scale metadata in Arrow result set processing for TIME, TIMESTAMP_NTZ, TIMESTAMP_LTZ, and TIMESTAMP_TZ columns (SNOW-3388299).[pandas] extra to allow installation of pandas 3.0.0 and later.x-amz-bucket-region response header on each redirect. Redirects are capped at 5 hops.WorkloadIdentityCredential to authenticate via the injected service account credentials. OIDC backward compatibility is also supported.workload_identity_aws_use_outbound_token connection option (default false) to opt into AWS WIF JWT attestation via STS GetWebIdentityToken instead of the default SigV4 GetCallerIdentity method.CREATE VIEW db.schema.obj) on a session with no current schema would populate the connector's cached _schema/_database from the referenced object's namespace. This made get_current_schema() diverge from the server's CURRENT_SCHEMA() and mis-qualified Snowpark temp objects (SNOW-3665226).Dropped support for Python 3.9. The minimum supported version is now Python 3.10.
.so/.dylib/.dll into the install prefix, breaking downstream packagers (e.g. Homebrew) whose audits reject foreign-arch binaries.account_provided, account_with_region, account_org_provided, region_provided, host_provided). No hostname or account value is included. This is gated by the existing server-side CLIENT_TELEMETRY_ENABLED parameter and can additionally be disabled locally by setting SF_TELEMETRY_DISABLE_CONNECTION_SHAPE=true. The telemetry collection is time-boxed and will be removed in a future release.urllib3 to 2.7.0Fixed write_pandas temp stage name collisions (SNOW-3481510). The old PRNG could produce identical name sequences in forked processes (e.g. Notebook k
write_pandas temp stage name collisions (SNOW-3481510). The old PRNG could produce identical name sequences in forked processes (e.g. Notebook kernels), causing CREATE TEMPORARY STAGE to fail with "Object already exists"._is_prefix_equal() compared url1's port against itself instead of url2's port, allowing an attacker to redirect credentials to a different port on the same hostname. Also fixed the default port fallback to use int instead of str for correct comparison when one URL omits the port.executemany with paramstyle="pyformat" to correctly locate the VALUES clause using a balanced-parentheses parser instead of a greedy regex. This fixes incorrect behaviour with nested function calls such as SQLAlchemy @compiles VARIANT patterns (e.g. PARSE_JSON(%(col)s)) and subquery-form INSERTs (SNOW-298756).SNOWFLAKE_ENABLE_AWS_WIF_OUTBOUND_TOKEN environment variable to true. Note: This environment variable will be removed in a future release.SNOWFLAKE_ENABLE_CUSTOM_REVOCATION_ERRORS environment variable is now a no-op.SNOWFLAKE_RUNNING_INSIDE_SPCS environment variable, trim whitespace, and remove configurable token path.metadata.google.internal instead of the IPv4 link-local address, so it works on IPv6-only GCP VMs.write_pandas() with auto_create_table=False and overwrite=True would execute CREATE TABLE IF NOT EXISTS, which required unnecessary OWNERSHIP privilege on the table. Now only TRUNCATE TABLE is executed in this case. Note: users who relied on the table being implicitly created despite auto_create_table=False should set auto_create_table=True instead.account connection parameter so malformed identifiers (for example path-like values or labels outside letters, digits, _, and -) are rejected with ProgrammingError before login (SNOW-1902886).Bump the lower boundary of cryptography to 46.0.5 due to CVE-2026-26007.
Metadata header to use lowercase "true" instead of "True", which caused 400 errors during Azure Workload Identity Federation authentication.crl_download_max_size to be 20MB instead of 200MB, as the previous value was set too high and could cause out-of-memory issues.SF_SKIP_WARNING_FOR_READ_PERMISSIONS_ON_CONFIG_FILE to SF_SKIP_TOKEN_FILE_PERMISSIONS_VERIFICATION. The old variable is still supported but emits a deprecation warning.unsafe_skip_file_permissions_check flag not being respected when reading connections.toml.Deprecated support for custom revocation error classes in OCSP response cache deserialization. By default, only RevocationCheckError exceptions are de…
server_session_keep_alive in SnowflakeConnection skip checking for pending async queries, providing faster connection close times especially when many async queries are executed.RevocationCheckError exceptions are deserialized from OCSP cache. Custom exception classes can be temporarily enabled by setting the SNOWFLAKE_ENABLE_CUSTOM_REVOCATION_ERRORS environment variable to true or 1, but this support will be removed in a future release.urllib3 to 2.6.3force_microseconds_precision to cursor.fetch_arrow_all and cursor.fetch_pandas_all to avoid PyArrow schema incosistency between batches.New features:
~ in token_file_path, so a path such as ~/.snowflake/pat.token is read from the home directory when loading a PAT, OAuth, or OIDC bearer token. (snowflake-eng/drivers#3049)Changes:
SQL_DESC_TYPE_NAME for FILE columns to report FILE. SQLDescribeCol stays SQL_VARCHAR, including when ODBC_USE_CUSTOM_SQL_DATA_TYPES is enabled. (snowflake-eng/drivers#2833)#. (snowflake-eng/drivers#2963)SQLBindCol and SQLGetData with SQL_C_DEFAULT on INTERVAL result columns to return a SQL_INTERVAL_STRUCT instead of the ANSI interval literal string, failing with 22015 when the leading field exceeds the interval leading precision (2 digits by default) and with HY090 when a nonzero buffer length is smaller than the struct; binding SQL_C_CHAR still returns the string. (snowflake-eng/drivers#3020)Bug fixes:
X-Snowflake-Service request header, echoing the SERVICE_NAME session parameter on queries so the proxy routes requests to the correct HTAP cluster. (snowflake-eng/drivers#3068)SQL_DESC_LOCAL_TYPE_NAME on those columns now matches the 3.x driver (SSMALLINT, USMALLINT, SINTEGER), and unsigned USER_DATA_TYPE reports display size 5. Catalog wide-character columns report octet length as four bytes per character, and SQLSpecialColumns COLUMN_NAME reports character length 255. (snowflake-eng/drivers#2760)SQLGetEnvAttr and SQLSetEnvAttr of an unrecognized environment attribute returning HYC00 instead of HY092. (snowflake-eng/drivers#3179)SF_DRIVER_NAME, and an upgrade keeps that registration unless SF_DRIVER_NAME names a different driver, which replaces the old registration. (snowflake-eng/drivers#2964)snowflake DSN stay registered, while an upgrade from a driver version released before this fix registers SF_DRIVER_NAME or the default Snowflake ODBC name and leaves a different custom name from that version registered. (snowflake-eng/drivers#2964)/opt/snowflake/snowflakeodbc/lib/libsfodbc.dylib. (snowflake-eng/drivers#3051)/Library/ODBC/odbcinst.ini as well as for the installing user. (snowflake-eng/drivers#3055)SQLGetEnvAttr(SQL_ATTR_DRIVER_UNICODE_TYPE) returning an unsupported-attribute error, so iODBC 3.52.14 and later converts wide strings to the driver's UTF-16 or UTF-32 encoding. (snowflake-eng/drivers#3179)SQLBindCol arrays bound as SQL_C_DEFAULT with a zero BufferLength writing every fetched row into the first array element instead of advancing by the size of the default C type. (snowflake-eng/drivers#3027)SQLBindParameter with SQL_C_DEFAULT and a SQL_INTERVAL_* parameter type reading the bound SQL_INTERVAL_STRUCT as a character string instead of as the matching SQL_C_INTERVAL_* type. (snowflake-eng/drivers#3020)SQLPutData rejecting character and binary data bound as SQL_C_DEFAULT with HY019 when it was sent in more than one piece. (snowflake-eng/drivers#3027)SQLBindParameter with SQL_C_DEFAULT and SQL_REAL reading the bound value as an 8-byte double instead of a 4-byte SQL_C_FLOAT, which inserted wrong values and misread parameter arrays. (snowflake-eng/drivers#3010)SQL_C_DEFAULT on SQL_WCHAR, SQL_WVARCHAR, and SQL_WLONGVARCHAR treating data as narrow characters instead of SQL_C_WCHAR, which truncated or corrupted bound wide-string parameters and returned catalog function wide-character columns as narrow strings. (snowflake-eng/drivers#3013)SQLColumns and SQLGetTypeInfo NUM_PREC_RADIX reporting INTEGER instead of the SMALLINT type the ODBC specification defines. (snowflake-eng/drivers#2760)SQLTables and SQLColumns so that, with CLIENT_METADATA_REQUEST_USE_CONNECTION_CTX enabled, a non-NULL catalog argument such as % is no longer limited to the current schema, and an unset session database or schema no longer makes the call return no rows. (snowflake-eng/drivers#2862)DRIVER_NAME removes that registration when this version or a later one installed it and leaves any other Snowflake ODBC registration in place, while an upgrade from an earlier installer still removes the default Snowflake ODBC registration unless that version is uninstalled with its original DRIVER_NAME first. (snowflake-eng/drivers#2963)SFGetFipsStatus before connecting. (snowflake-eng/drivers#2837)validate_default_parameters=true not failing login when the specified database, schema, or warehouse does not exist. (snowflake-eng/drivers#2974)sf.odbc.ini sets DriverManagerEncoding=UTF-32; the driver now uses UTF-16 and logs a warning. (snowflake-eng/drivers#2961)Added SnowflakeCursor.stats property to expose granular DML statistics (rows inserted, deleted, updated, and duplicates) for operations like CTAS wher
SnowflakeCursor.stats property to expose granular DML statistics (rows inserted, deleted, updated, and duplicates) for operations like CTAS where rowcount is insufficient.SPCS_TOKEN) into login requests when present in SPCS containers.Changes:
SQLDriverConnect so unrecognized connection-string keywords are reported with a local 01S00 warning and are not forwarded to the server. Recognized session parameters are sent to the server.TRACING keyword in a DSN or connection string to post a SQLSTATE 01000 deprecation warning on connect that names LogLevel in sf.odbc.ini; the keyword still has no effect.CLIENT_SESSION_KEEP_ALIVE and CLIENT_PREFETCH_THREADS unless the connection sets them.validate_default_parameters value, true or false, as CLIENT_VALIDATE_DEFAULT_PARAMETERS, and to leave that parameter off the login request when it is unset.Bug fixes:
ConfigDSN (including Add-OdbcDsn) failing with "Failed to save DSN configuration." or a FALSE return; saving now leaves a stored TRACING value in place.SQLDriverConnect so ODBC_QUERY_RESULT_FORMAT is sent to the server and selects the session query result format.TIMEZONE overrides so a TIMEZONE set on a statement is sent on the query request.SQLTables so the SQL_ALL_CATALOGS and SQL_ALL_SCHEMAS listings return the account's catalogs or schemas when TableType is null or any other value, instead of an empty result.Relaxed pandas dependency requirements for Python below 3.12.
Pin lower versions of dependencies to oldest version without vulnerabilities.
SNOWFLAKE_AUTH_FORCE_SERVER environment variable to force the use of the local-listening server when using the externalbrowser auth method.
oauth_socket_uri connection parameter allowing to separate server and redirect URIs for local OAuth server.New features:
CRL_ALLOW_NO_CRL, CRL_DISK_CACHING, CRL_MEMORY_CACHING, and CRL_DOWNLOAD_TIMEOUT as aliases for the matching 4.x CRL settings. CRL_DOWNLOAD_TIMEOUT=0 applies the 4.x crl_http_timeout default of 10 seconds rather than 3.x's 120.SFGetFipsStatus so an application can check, before connecting, whether this process loaded the FIPS driver.APPLICATION_PATH, matching the 3.x driver.SQL_SF_TIMESTAMP_LTZ, SQL_SF_TIMESTAMP_TZ, SQL_SF_TIMESTAMP_NTZ, SQL_SF_ARRAY, SQL_SF_OBJECT, SQL_SF_VARIANT) to the public sf_odbc.h, so applications written against the 3.x header compile without redefining them.MapToLongVarchar connection parameter so CHAR and VARCHAR columns whose size exceeds the threshold are reported as SQL_LONGVARCHAR. Unset and negative values keep SQL_VARCHAR.DEFAULT_VARCHAR_SIZE and DEFAULT_BINARY_SIZE connection parameters so max-length VARCHAR and BINARY columns report the configured sizes. Unset and negative values keep the session maximum.PUT_COMPRESSLV connection parameter so PUT AUTO_COMPRESS can select gzip compression levels 0–9. Unset and out-of-range values keep gzip level 6.PUT_TEMPDIR connection parameter so PUT AUTO_COMPRESS can write gzip tempfiles to a caller-supplied directory. Unset and empty values keep the process temp directory. Nested directories are created.SQL_SF_TIMESTAMP_LTZ, SQL_SF_TIMESTAMP_TZ, SQL_SF_TIMESTAMP_NTZ, SQL_SF_ARRAY, SQL_SF_OBJECT, SQL_SF_VARIANT) to the public sf_odbc.h, so applications written against the 3.x header compile without redefining them.SQL_SF_VECTOR (2006) to the public sf_odbc.h so applications can name the VECTOR type that SQLGetTypeInfo already reports. Binding VECTOR remains unsupported.sf.odbc.ini logging configuration: each SQLConnect/SQLDriverConnect/SQLBrowseConnect now re-resolves LogLevel/LogPath (and the troubleshooting environment variables) and applies any change in place before the connection proceeds, so editing the ini or dropping in a new one takes effect on the next connect with no process restart. An unchanged configuration is detected by a cheap fingerprint and skipped without reading the file; DriverManagerEncoding stays negotiated once per process and DSN LogLevel/LogPath/TRACING remain ignored.wif_host connection parameter so Workload Identity Federation can target a specific AWS STS endpoint; unset keeps the regional default.PAT_WITH_EXTERNAL_SESSION authentication (authenticator=PAT_WITH_EXTERNAL_SESSION with external_session_id) so a programmatic access token can reuse a server-side session.Changes:
CRL_CHECK and CRL_ADVISORY to be accepted and ignored, posting SQLSTATE 01000 on connect. Set CRL_MODE to ENABLED for fail-close revocation checking or ADVISORY for fail-open.Bug fixes:
SQLColAttribute to report ARRAY/STRUCT/VARIANT for SQL_DESC_TYPE_NAME and SQL_DESC_LOCAL_TYPE_NAME instead of always returning VARCHAR.SQL_DESC_TYPE_NAME so TIMESTAMP_NTZ, TIMESTAMP_LTZ, and TIMESTAMP_TZ are reported separately. SQLDescribeCol stays SQL_TYPE_TIMESTAMP (93).SQLBindParameter and IPD SQLSetDescRec to return SQLSTATE HYC00 (optional feature not implemented) for the Snowflake vendor type codes that SQLGetTypeInfo reports but the driver cannot bind: SQL_SF_ARRAY, SQL_SF_OBJECT, SQL_SF_VARIANT, and SQL_SF_VECTOR. SQLBindParameter previously returned HY004; IPD SQLSetDescRec previously returned HY021. Bind semi-structured values as SQL_VARCHAR and coerce in the SQL with PARSE_JSON(?), TO_ARRAY(?), or TO_OBJECT(?). A type code the catalog does not report still returns HY004 from SQLBindParameter.@ unless the authority is a dotted host or bracketed IPv6 with a numeric port; query strings and fragments are also stripped from those parameters.COPY INTO <location> so SQLNumResultCols and SQLFetch expose the unload result rows instead of reporting no result set.sf.odbc.ini discovery so a candidate path that is a directory (or otherwise not a regular file) is skipped instead of aborting the search; a SF_ODBC_INI pointing at a directory no longer hides the standard per-user sf.odbc.ini.Added support for checking certificates revocation using revocation lists (CRLs)
CERT_REVOCATION_CHECK_MODE to CLIENT_ENVIRONMENTworkload_identity_impersonation_path parameter to support service account impersonation for Workload Identity Federation on GCP and AWS workloads onlyget_results_from_sfqid when using DictCursor and executing multiple statements at onceoauth_credentials_in_body parameter supporting an option to send the oauth client credentials in the request bodyECONNRESET errorbotocore and boto3 dependencies by setting SNOWFLAKE_NO_BOTO environment variable during installationOauth authenticator back to DatabaseErrorConfigSourceError with detailed permission information, preventing potential credential tampering.SnowflakeConnection.cursor(cursor_class) to match the type of cursor_classfetchone, fetchmany, fetchall
DictCursor is no longer a subclass of SnowflakeCursor; use SnowflakeCursorBase as a superclass of both.AWS_DEFAULT_REGION variable instead of AWS_REGION for WORKLOAD_IDENTITY authenticatorocsp_root_certs_dict_lock_timeout connection parameter to set the timeout (in seconds) for acquiring the lock on the OCSP root certs dictionary. Default value for this parameter is -1 which indicates no timeout.Breaking changes:
CatalogName is no longer replaced with the current database by default. Set UseCurrentCatalog=true (or enable CLIENT_METADATA_REQUEST_USE_CONNECTION_CTX) to restore that substitution. Unconstrained NULL-catalog searches issue account-wide SHOW statements.SQLColumns BUFFER_LENGTH for DATE/TIME from COLUMN_SIZE (10 / 18 for TIME(9)) to 6 (sizeof(SQL_DATE_STRUCT) / sizeof(SQL_TIME_STRUCT)); query-result SQLColAttribute octet length for DATE/TIME remains 6.SQLColumns BUFFER_LENGTH for NUMBER/DECIMAL to return precision + 2 (ODBC transfer octet length); query-result SQLColAttribute octet/display for NUMBER remains 136.SQLColumns SQL_DATA_TYPE for DATE/TIME/TIMESTAMP to the verbose SQL_DATETIME (9) with the subtype in SQL_DATETIME_SUB; DATA_TYPE still returns the concise type (91 / 92 / 93).SQLColumns COLUMN_SIZE for TIMESTAMP* from a fixed 35 to 20 + scale (19 when scale is 0).SQLColumns BUFFER_LENGTH for TIMESTAMP* from 35 to 16 (sizeof(SQL_TIMESTAMP_STRUCT)).SQLColumns COLUMN_SIZE and BUFFER_LENGTH for VARIANT/OBJECT/ARRAY to follow VARCHAR_AND_BINARY_MAX_SIZE_IN_RESULT instead of SHOW COLUMNS' 128 MB length.SQL_C_CHAR conversion of DECIMAL/NUMERIC to return SQL_ERROR (22003) when whole digits do not fit in the buffer; previously returned SQL_SUCCESS_WITH_INFO with truncated digits.SQL_C_BINARY conversion of DECIMAL/NUMERIC/DECFLOAT to return SQL_ERROR (22003) when the buffer is smaller than sizeof(SQL_NUMERIC_STRUCT); previously BufferLength was ignored, risking buffer overflows.FLOAT/DOUBLE to integer and SQL_C_BIT targets to return SQL_ERROR (22003); previously returned SQL_SUCCESS_WITH_INFO with a silent 0.SQL_C_INTERVAL_* types: values that exceed the default precision of 2 now return SQL_ERROR (22015), and SQL_DESC_DATETIME_INTERVAL_PRECISION is respected; previously neither was enforced.SQLRowCount with a NULL RowCountPtr to return SQL_ERROR (HY009); previously returned SQL_SUCCESS.SQLBindParameter to return SQL_ERROR (HY104) for a negative DecimalDigits value; previously accepted negative scale silently.SQL_C_BINARY to return SQL_ERROR (22003) instead of silently clamping.SQL_C_NUMERIC to VARCHAR conversion to apply the scale from SQL_NUMERIC_STRUCT (e.g. magnitude 12345 with scale 2 produces "123.45"); previously scale was ignored.DBL_MIN to preserve the value; previously stored as 0.0.TIME to SQL_C_CHAR/SQL_C_WCHAR buffer-too-small handling: a buffer that cannot hold the base time (HH:MM:SS) now returns SQL_ERROR (22003); a buffer that truncates only fractional seconds returns SQL_SUCCESS_WITH_INFO (01004).DATE to SQL_C_BINARY conversion with an undersized buffer to return SQL_ERROR (22003); previously returned SQL_SUCCESS with truncated data.SQLGetFunctions to report SQL_API_SQLCANCELHANDLE as supported; previously returned SQL_FALSE despite the function being exported.SQL_C_INTERVAL_* types. Truncation returns SQL_SUCCESS_WITH_INFO (01S07) and interval-field overflow returns SQL_ERROR (22015).SQL_C_INTERVAL_SECOND with fractional seconds bound to exact-numeric SQL types to truncate the fraction and succeed; previously returned SQL_ERROR (22015).SQLCancel during data-at-execution to discard all accumulated SQLPutData data so a re-entered sequence starts fresh; previously accumulated data from an interrupted sequence could be concatenated into the next sequence.SQLForeignKeys to resolve a NULL FK catalog and schema from the connection context; previously returned an empty result set unless CLIENT_METADATA_REQUEST_USE_CONNECTION_CTX was enabled.SQLSetConnectAttr(SQL_ATTR_LOGIN_TIMEOUT) called after connect to return SQL_ERROR (HY011); previously returned SQL_SUCCESS as a silent no-op.SQLSetStmtAttr(SQL_ATTR_CURSOR_TYPE) for unsupported cursor types to substitute SQL_CURSOR_FORWARD_ONLY and return SQL_SUCCESS_WITH_INFO (01S02); previously accepted any value silently.[Snowflake][Support] to [Snowflake][Snowflake ODBC Driver], matching the ODBC spec format [vendor][ODBC-component-identifier].SQLGetInfo(SQL_DRIVER_NAME) to return the loaded driver library's file name (e.g. libsfodbc.so, libsfodbc.dylib, sfodbc.dll) as the ODBC specification prescribes; previously returned the fixed string Snowflake.SQLCancelHandle(SQL_HANDLE_DBC) to return SQL_ERROR (HY010) when an associated statement is asynchronously executing or mid data-at-execution; previously returned SQL_SUCCESS as a no-op.SQL_C_CHAR/SQL_C_WCHAR to return normalized scientific notation (e.g. 1.2e200 instead of 12e199).> to >= and added support for the arrayBindSupported and CLIENT_STAGE_ARRAY_BINDING_THRESHOLD session parameters; previously the server-provided arrayBindSupported flag was ignored.SQL_SF_CONN_ATTR_PRIV_KEY (raw EVP_PKEY*) connection attribute; use PRIV_KEY_CONTENT, PRIV_KEY_BASE64, or PRIV_KEY_FILE instead.http:// is allowed only for loopback addresses.HY000 to 28000 (invalid authorization specification).WORKLOAD_IDENTITY_PROVIDER) unless AUTHENTICATOR=WORKLOAD_IDENTITY; previously silently ignored them with any other authenticator.WORKLOAD_IDENTITY_IMPERSONATION_PATH with IDENTITY_PROVIDER=OIDC to return a connection error; previously silently ignored the combination.REMARKS and SQLColumns COLUMN_DEF to return SQL_NULL_DATA when absent instead of an empty string, and SQLTables REMARKS now surfaces a SHOW OBJECTS comment when present.SQL_SF_STMT_ATTR_LAST_QUERY_ID to be read-only and fully populated after every statement execution; changed SQL_SF_STMT_ATTR_MULTI_STATEMENT_COUNT to support get/set with a default of -1 (auto-detect); neither attribute was fully implemented in ODBC 3.x.SQLGetDiagRec / SQLGetDiagField message text to append an internal error trace by default; set ErrorTraceEnabled=false in sf.odbc.ini to restore 3.x-shaped message text.SQL_NO_DATA, length-only queries return SQL_SUCCESS, and required SUCCESS_WITH_INFO / HY012 diagnostics are posted correctly.SQLDisconnect and SQLFreeStmt(SQL_DROP) now invalidate all associated handles through iODBC's alloc table so a subsequent SQLFreeHandle on an orphaned handle returns SQL_INVALID_HANDLE instead of SQL_SUCCESS.SQLGetDescField and SQLSetDescField called during SQL_NEED_DATA state under iODBC to return SQL_ERROR (HY010); previously returned SQL_SUCCESS.HY090, HY010, HY017, HY092, 07009, HY008); previously returned vendor HY000 or ODBC 2.x aliases.SQL_C_WCHAR fetch encoding under iODBC to use a uniform UTF-32 width driven by DriverManagerEncoding in sf.odbc.ini; previously the code-unit width varied by Snowflake source column type.HTTP_PROXY / HTTPS_PROXY / NO_PROXY environment variables are ignored unless USE_PROXY_ENV=true (alias PROXYWITHENV); explicit PROXY= / NO_PROXY= connection parameters still apply. ODBC 3.x fell back to those environment variables when no DSN/simba.snowflake.ini proxy was set. PROXYWITHENV no longer writes the connection's PROXY value into the process environment.SF_GLOBAL_SSL_VERSION (and TLS settings) below TLS 1.2 to fail the connection; previously logged a deprecation warning and still negotiated.New features:
MapToLongVarchar connection parameter so CHAR and VARCHAR columns whose size exceeds the threshold are reported as SQL_LONGVARCHAR. Unset and negative values keep SQL_VARCHAR.DEFAULT_VARCHAR_SIZE and DEFAULT_BINARY_SIZE connection parameters so max-length VARCHAR and BINARY columns report the configured sizes. Unset and negative values keep the session maximum.PUT_COMPRESSLV connection parameter so PUT AUTO_COMPRESS can select gzip compression levels 0–9. Unset and out-of-range values keep gzip level 6.PUT_TEMPDIR connection parameter so PUT AUTO_COMPRESS can write gzip tempfiles to a caller-supplied directory. Unset and empty values keep the process temp directory. Nested directories are created.MaxHttpRetries connection-string key as an alias of retry_max_attempts.PUT_GET_MAX_ATTEMPTS connection parameter for the shared PUT/GET attempt limit, and accepted the 3.x PUT_MAXRETRIES / GET_MAXRETRIES spellings as aliases that warn (01000) on use.UseCurrentCatalog connection parameter (default false), matching the 3.x DSN key. When true, a NULL CatalogName on SQLTables, SQLColumns, SQLPrimaryKeys, SQLForeignKeys, SQLProcedures, and SQLProcedureColumns is the current database.INTERVAL YEAR TO MONTH and INTERVAL DAY TO SECOND result support: SQL_C_CHAR/SQL_C_WCHAR fetch returns the canonical ANSI literal ([-]Y-MM, [-]D HH:MM:SS[.f]), same-family SQL_C_INTERVAL_* targets receive the parsed interval struct, and scalar numeric targets receive total months or total whole seconds (reporting 01S07 when sub-second precision is dropped).AZURE_CLIENT_ID, AZURE_TENANT_ID, and AZURE_FEDERATED_TOKEN_FILE into a pod and the projected token file exists on disk, WORKLOAD_IDENTITY_PROVIDER=AZURE exchanges that federated token for an Entra ID access token directly. WORKLOAD_IDENTITY_IMPERSONATION_PATH is not supported in this environment.WORKLOAD_IDENTITY_AWS_USE_OUTBOUND_TOKEN connection parameter for AWS Workload Identity Federation. When set to true, attestation uses outbound STS GetWebIdentityToken instead of the default pre-signed GetCallerIdentity token; the connection parameter takes precedence over SNOWFLAKE_ENABLE_AWS_WIF_OUTBOUND_TOKEN.INCLUDE_RETRY_REASON (default true) so retried query requests send retryReason (the HTTP status that triggered the retry, or 0 for transport failures) alongside retryCount.SQLFreeConnect (ODBC 2.x) as a thin wrapper around SQLFreeHandle(SQL_HANDLE_DBC, …) for direct-link and ODBC 2.x applications that bypass the Driver Manager.SQLFreeEnv (ODBC 2.x) as a thin wrapper around SQLFreeHandle(SQL_HANDLE_ENV, …) for direct-link and ODBC 2.x applications that bypass the Driver Manager.TOKEN_FILE_PATH connection parameter that loads a PAT, legacy OAuth, or OIDC bearer token from a file; when both TOKEN and TOKEN_FILE_PATH are set, the file contents are used.[1,2,3]) via SQL_C_CHAR, SQL_C_WCHAR, and SQL_C_BINARY; SQLDescribeCol reports SQL_VARCHAR and SQL_DESC_TYPE_NAME returns VECTOR.ALTER USER ADD PROGRAMMATIC ACCESS TOKEN result-set support: the generated token is returned as a result row; ODBC 3.x returned an invalid cursor state for this DDL.SQLTransact (ODBC 2.x) as a mapping to SQLEndTran, restoring compatibility with ODBC 2.x applications that bypass the Driver Manager.TIME to SQL_C_BINARY conversion support.SQL_C_BINARY as SQLBindParameter source type for SQL_DECIMAL, SQL_NUMERIC, SQL_TIME, and SQL_TIMESTAMP when the buffer length matches the SQL data length.~/ or ~ in the source path is expanded to the user's home directory.GET operations.SQLGetDiagField(SQL_DIAG_SERVER_NAME) population: returns the connected server name instead of an empty string.SQLGetDiagField(SQL_DIAG_DYNAMIC_FUNCTION) / SQL_DIAG_DYNAMIC_FUNCTION_CODE population after statement execution (e.g. SELECT → 85 / SQL_DIAG_SELECT_CURSOR); previously returned empty / 0.SQL_C_BINARY to VARCHAR binding: binary bytes are hex-encoded before send (e.g. 0xDEADBEEF → "deadbeef"); ODBC 3.x forwarded raw bytes and the server rejected the request with HTTP 400.SQL_SF_TIMESTAMP_TZ parameter binding for SQL_C_TYPE_TIMESTAMP (stored at UTC) and SQL_C_CHAR/SQL_C_WCHAR strings with a +/-HH:MM offset.SQLBindParameter support for binding to SQL_INTERVAL_* parameters from same-family SQL_C_INTERVAL_*, SQL_C_CHAR/SQL_C_WCHAR, and exact-numeric C sources (approximate-numeric sources are rejected with SQL_ERROR 07006).SQL_SF_CONN_ATTR_APPLICATION as a pre-connect get/set attribute that forwards the application name to the server.SQLGetFunctions reporting of SQL_API_SQLSETSCROLLOPTIONS and SQL_API_SQLPARAMOPTIONS as supported under iODBC.Changes:
SQLRowCount after a SELECT capped by ROWS_PER_RESULTSET so it reports the number of rows in the result payload, matching ODBC 3.x.LogLevel and LogPath to be accepted and ignored, posting SQLSTATE 01000 on connect. Each warning names the sf.odbc.ini key that configures the same behavior (for example LogFileCount points at LogMaxCount).INJECT_CURL_TIMEOUT, INJECT_INCIDENT1, CURL_NO_IDLE_CHECK, DisableStageBindFallback, StageBindMaxFileSize, StageBindThreshold) to be accepted and ignored, posting SQLSTATE 01000 on connect. These knobs have no replacement; remove them from the DSN.CPTimeout connection-string keyword to be accepted and ignored instead of treated as an unknown session parameter.DisableOCSPCheck and OCSP_FAIL_OPEN to be accepted and ignored, posting SQLSTATE 01000 on connect. Certificate revocation uses CRL; set CRL_MODE to DISABLED, ENABLED, or ADVISORY.CLIENT_STORE_TEMPORARY_CREDENTIAL to default to true when the caller has not set it. An explicit value always wins.TRANSLATE to be accepted and ignored, posting SQLSTATE 01000 on connect. Character-set translation DLLs are not supported.SQLGetInfo(SQL_ASYNC_DBC_FUNCTIONS) to report SQL_ASYNC_DBC_NOT_CAPABLE, matching the driver's rejection of SQL_ATTR_ASYNC_DBC_FUNCTIONS_ENABLE; on Windows the Driver Manager now rejects that attribute with HY114 before the driver sees it, instead of the driver's HY092.SQLSetConnectAttr(SQL_ATTR_ASYNC_DBC_FUNCTIONS_ENABLE) so unixODBC and iODBC accept SQL_ASYNC_DBC_ENABLE_OFF and read it back, while enabling still fails with HYC00; on Windows the Driver Manager rejects both ON and OFF with HY114 once GetInfo reports NOT_CAPABLE.SQLGetTypeInfo TIMESTAMP COLUMN_SIZE from 35 to 29 (the ODBC-standard display length). TIMESTAMP_LTZ, TIMESTAMP_NTZ, and TIMESTAMP_TZ remain 35. The 3.x ODBC_USE_STANDARD_TIMESTAMP_COLUMNSIZE connection parameter is not accepted.SQLDriverConnect to reject connection-string keywords it does not recognize with a local 01S00 warning (native error 17, "N invalid keys are found in the connection string: "); a keyword is recognized when the sf_core parameter registry resolves it or it names an ODBC structural keyword (DSN, DRIVER, FILEDSN, SAVEFILE). The connection still opens and the keyword is still forwarded to the server.Tracing(0-6) field from the Windows ODBC DSN setup dialog; driver logging uses sf.odbc.ini (LogLevel, LogPath) instead. Legacy TRACING values in a DSN or connection string are ignored.DSN and DRIVER to be accepted and ignored instead of forwarded as unknown session parameters.HY000 instead of the warning-class 01S00; affected cases are an invalid PORT, an unparseable connection string, and any invalid or missing non-credential parameter. 01S00 is defined by the ODBC specification as a SQL_SUCCESS_WITH_INFO warning meaning the connection opened anyway, so returning it on SQL_ERROR led applications that branch on the SQLSTATE class to treat a failed connection as a warning. ODBC 3.x returned 28000 (native error 20032) for these cases.PRIVATE_KEY / PRIVATE_KEY_FILE) and a missing bearer token (TOKEN / TOKEN_FILE_PATH) to report SQLSTATE 28000 instead of 01S00, matching ODBC 3.x and the SQLSTATE already reported when either parameter is named on its own.WORKLOAD_IDENTITY_PROVIDER, WORKLOAD_IDENTITY_ENTRA_RESOURCE, WORKLOAD_IDENTITY_IMPERSONATION_PATH, WORKLOAD_IDENTITY_AWS_USE_OUTBOUND_TOKEN) to report SQLSTATE 28000 rather than a general error, so a missing or invalid one is reported as the authentication failure it is. ODBC 3.x reported 28000 for a missing WORKLOAD_IDENTITY_PROVIDER.SQLGetDiagRec now reads Failed to login: Login error: <server text>, code: <code> on its first line, where the server sentence previously appeared only inside the error trace. SQLSTATE and native error code are unchanged.SQLBrowseConnect so an incomplete connection string returns SQL_NEED_DATA and keeps the handle available for further browse calls, matching the 3.x iterative protocol under iODBC.CLIENT_STORE_TEMPORARY_CREDENTIAL to true when the caller has not set it, matching ODBC 3.x token caching.SQLForeignKeys with SQL_ATTR_METADATA_ID=TRUE to return SQL_ERROR (HY009) for a NULL catalog, schema, or table pointer on either side.SQL_C_BINARY fetch of FLOAT/DOUBLE/REAL to return the native 8-byte IEEE 754 value instead of a 19-byte SQL_NUMERIC_STRUCT.SQL_BIT parameter binding so integer and SQL_C_NUMERIC sources accept only 0 and 1 (other magnitudes return 22003).SQL_C_CHAR/SQL_C_WCHAR binding of "Infinity", "-Infinity", and "NaN" to SQL_FLOAT/SQL_REAL/SQL_DOUBLE to forward the non-finite value instead of returning 22018.SQL_C_CHAR/SQL_C_WCHAR hex literals bound to SQL_BINARY so an odd-length hex string drops the leftover nibble and succeeds.SQLBindParameter with Snowflake vendor TIMESTAMP codes (2000 / 2001 / 2002) to store those codes on the IPD so SQLDescribeParam returns them as bound.SQLSetStmtAttr(SQL_ROWSET_SIZE, 0) to return SQL_ERROR (HY024) instead of storing 0 or coercing to 1.source_compression / target_compression tokens to lowercase (e.g. gzip), matching ODBC 3.x.FNAME header, matching ODBC 3.x.PARALLEL value; result rows keep their original file order.TOKEN_FILE_PATH to report SQLSTATE 28000 instead of 01S00.SQLCancel to cancel through the statement's operation handle so the executing path aborts the query on the server; SQLCancel returns once the cancel is signaled, and the canceled call still reports HY008 after the abort is issued.HYT00 when no server SQLSTATE is present.SQL_C_CHAR/SQL_C_WCHAR fetch of +/-Infinity FLOAT/REAL to return INFINITY / -INFINITY instead of inf / -inf.TIMESTAMP_TZ Arrow fetch to reject a timezone offset outside the valid biased range 0 through 2880.TIMESTAMP_TZ Arrow fetch to reject a flat Int64 physical type instead of decoding it as UTC with offset 0.Snowflake ODBC; update DSN Driver= entries that reference SnowflakeDSIIDriver (custom names still configurable via DRIVER_NAME= / SF_DRIVER_NAME).snowflake-odbc-<version>.<arch>.<extension> with unified architectures aarch64 / x86_64 / x86_32 / universal.SQLGetInfo(SQL_DRIVER_VER) to return the zero-padded MM.mm.bbbb format (e.g. 04.00.0000) per the ODBC specification instead of the unpadded semver string (e.g. 3.16.0).sf.odbc.ini (searched in $SF_ODBC_INI, ~/.config/snowflake/sf.odbc.ini, ~/.snowflake/sf.odbc.ini, and /opt/snowflake/snowflakeodbc/sf.odbc.ini); update deployment scripts that write simba.snowflake.ini.PASSCODEINPASSWORD connection parameter to accept true and 1 in addition to on.QUERY_TAG connection parameter to be applied as a server session parameter at connection time.credential_cache_v1.json to credential_cache_v2.json (same lookup order: $SF_TEMPORARY_CREDENTIAL_CACHE_DIR, $XDG_CACHE_HOME/snowflake, $HOME/.cache/snowflake). Tokens cached by ODBC 3.x are not read by 4.x, so expect one extra authentication after upgrading.0600) permissions by default; set UNSAFE_FILE_WRITE=true to use the process umask.PRIV_KEY_FILE / private_key_file reads on Unix to require owner-only permissions (mode 0600); overly permissive key files now fail unless UNSAFE_SKIP_CONFIG_FILE_PERMISSIONS_CHECK=true.AUTHENTICATOR=WORKLOAD_IDENTITY to reject hosts outside the recognized Snowflake suffixes (snowflakecomputing.com / .cn / .mil) before fetching cloud credentials; extend the list only via the SNOWFLAKE_WIF_ALLOWED_HOST_SUFFIXES environment variable.KeyWrappingMetadata.EncryptionLibrary metadata to "Rust(OpenSSL)".ACCOUNT / SERVER / PORT characters to SQLSTATE 01S00 (native error 0); ODBC 3.x returned 28000 (native error 20032).CLIENT_SESSION_KEEP_ALIVE_HEARTBEAT_FREQUENCY to be sent as a server session parameter during login (queryable via SHOW PARAMETERS) in addition to configuring the client heartbeat interval; ODBC 3.x used it client-side only.Bug fixes:
HH:MM:SS instead of nanoseconds since midnight, matching ODBC 3.x bulk binding.SQLPrepare on statements the server declines to describe, such as ALTER SESSION and COMMIT, which previously failed with SQLSTATE 0A000 and native error 7 (statement not preparable); prepare now succeeds and the statement runs on SQLExecute, matching ODBC 3.x.SQL_DESC_TYPE_NAME reports GEOGRAPHY or GEOMETRY while preserving the text or binary concise type selected by the output format.SQLExecute/SQLExecDirect to return SQLSTATE 07S01 (Invalid use of default parameter) when any bound parameter has StrLen_or_IndPtr = SQL_DEFAULT_PARAM (-5); previously the driver returned HY000 where the ODBC spec requires 07S01.SQL_TINYINT parameter binds so a value outside the -128 to 255 range is rejected at execute with SQLSTATE 22003 (Numeric value out of range) instead of being sent to the server unchecked.SQL_NULL_DATA indicator so execute reports SQLSTATE HY009 (Invalid use of null pointer) instead of the general error HY000.SQLExecDirect so lowering the application parameter descriptor SQL_DESC_COUNT after extra parameters were bound no longer fails a statement that uses fewer placeholders.SQLBindParameter so an unsupported C-to-SQL conversion is rejected at bind with SQLSTATE 07006, matching ODBC 3.x.SQL_C_DEFAULT parameter binds for DECIMAL, integer, DATE, TIME, and TIMESTAMP so execute infers the ODBC default C type instead of failing with SQLSTATE 07006.SQL_PARAM_OUTPUT binds so they are omitted from execute-time JSON/CSV encoding instead of failing with SQLSTATE HY009 when the output buffer pointer is null.SQL_C_CHAR or SQL_C_WCHAR so a non-numeric literal returns SQLSTATE 22018 instead of 07006.SQLPrepare + SQLExecute for PUT/GET file-transfer commands, which previously failed with server error 000007 (statement not preparable) because prepare issued a describeOnly request the server rejects; prepare now skips the describe and the transfer runs on execute.SQLGetTypeInfo to return type information matching the application's configured ODBC version.SQL_WVARCHAR metadata consistently, including after SQLPrimaryKeys and SQLForeignKeys results are installed on the statement; SQLStatistics ASC_OR_DESC reports SQL_WCHAR.SQLDriverConnect failures on Windows x64 Azure during JWT login.SQL_C_DEFAULT on catalog SMALLINT and INTEGER columns so SQLGetTypeInfo and SQLColumns return binary integers instead of failing with SQLSTATE 22003.MINUTE TO SECOND, SECOND, and truncated DAY TO SECOND fractions).SQLColumns sizes for GEOGRAPHY and GEOMETRY to follow the session VARCHAR maximum.SQL_ATTR_ROW_BIND_TYPE strides so length, indicator, character, wide-character, and fixed-width value writes do not crash.SQL_C_CHAR/SQL_C_WCHAR binds to TIMESTAMP_TZ columns rejecting ISO8601 strings with a T date-time separator (e.g. "2017-11-30T18:17:05.123456789+08:00"); the parser now accepts both the space and T separator variants.SQLGetDiagField return codes for three edge cases: a record field requested with RecNumber=0 now returns SQL_ERROR instead of SQL_NO_DATA, a header field requested with a positive RecNumber now returns SQL_SUCCESS instead of SQL_NO_DATA, and a negative BufferLength for a string field now returns SQL_ERROR.SYSTEM$BIND stage is disabled, instead of failing the statement.0600 and then using it anyway; such a file is now reported and left unused.CLIENT_SESSION_KEEP_ALIVE_HEARTBEAT_FREQUENCY falls outside the accepted range; the value is now clamped before login.Driver=-only connect with no other connection-string attributes to load the default connections.toml profile.SQLGetConnectAttr and decimal-as-int conversion to honor typed values returned by the server (e.g. AUTOCOMMIT after ALTER SESSION).FILE column failing with Unsupported column type.MAP column failing with Unsupported column type.SQLColumns and SQLProcedureColumns CHAR_OCTET_LENGTH for VARIANT/OBJECT/ARRAY to report the session VARCHAR max (equal to BUFFER_LENGTH) instead of NULL.SQLColumns COLUMN_SIZE and BUFFER_LENGTH for unrecognized Snowflake types such as GEOGRAPHY/GEOMETRY to report the varchar metrics implied by their SQL_VARCHAR DATA_TYPE instead of NULL.SQLProcedureColumns TYPE_NAME for unsupported types such as GEOGRAPHY/GEOMETRY to report the Snowflake type name while DATA_TYPE remains SQL_VARCHAR.SQLColumns and SQLProcedureColumns CHAR_OCTET_LENGTH for unsupported types such as GEOGRAPHY/GEOMETRY to report a byte length instead of NULL, matching the SQL_VARCHAR they report as DATA_TYPE.SQLGetTypeInfo string result columns (TYPE_NAME, LITERAL_PREFIX/SUFFIX, CREATE_PARAMS, LOCAL_TYPE_NAME) to report SQL_WVARCHAR as the IRD concise type, matching SQLTables/SQLColumns.SQLGetTypeInfo INTERVAL_PRECISION to report SQL_SMALLINT as the IRD concise type, matching the ODBC spec and the reference driver; NUM_PREC_RADIX remains SQL_INTEGER.08001 instead of 28000.ACCOUNT value hanging until login timed out; it is now rejected immediately..part download file.SQLColumns and SQLProcedureColumns BUFFER_LENGTH and CHAR_OCTET_LENGTH for VARCHAR/TEXT to report Snowflake byteLength, falling back to 4× COLUMN_SIZE capped at the session VARCHAR byte maximum where no byteLength is available; query-result SQLColAttribute octet length is unchanged.GET downloads to leave no partial or corrupt file at the destination; the driver now writes to a .part temporary file and renames it on successful completion.SQLDisconnect to be thread-safe.SQLDisconnect to free child statement handles and explicitly allocated descriptors after a successful disconnect so a later SQLFreeHandle on those handles returns SQL_INVALID_HANDLE.SQLDisconnect to return SQL_ERROR (HY010) without disconnecting when a child statement is asynchronously executing or mid data-at-execution.SQLCancel on a statement with SQL_ATTR_ASYNC_ENABLE to actually interrupt the in-progress operation: subsequent polling returns SQL_ERROR (HY008) once the cancellation is acknowledged. Previously polling kept returning SQL_STILL_EXECUTING until the query completed naturally.SQLCancel and SQLCancelHandle to always return SQL_SUCCESS and post no diagnostics of their own, per the ODBC specification; previously they intermittently returned SQL_ERROR (HY008, native error 604) when the server-side abort raced the query's completion. Only the canceled function returns HY008.FLT_MAX, DBL_MAX) being incorrectly rejected with a numeric out-of-range error; they are now returned successfully.TIMESTAMP to SQL_C_CHAR/SQL_C_WCHAR conversion when the destination buffer was too small.SQLNumResultCols and SQLDescribeCol to return the correct column count and metadata after a prepared statement's cursor is closed - including the async SQLPrepare + SQLExecute + SQLCloseCursor path - instead of returning incorrect results or errors.SQLDescribeCol column size for SQL_DOUBLE / SQL_FLOAT to report 15 (decimal digit precision) instead of 53 (binary mantissa bits).SQLColumns NUM_PREC_RADIX for FLOAT/DOUBLE/REAL to return 10, matching decimal COLUMN_SIZE (query-result SQLColAttribute radix for DOUBLE remains 2).SQLFreeHandle(SQL_HANDLE_DESC) on an implicitly allocated descriptor to return SQL_ERROR (HY017) and leave the handle valid.SQL_C_CHAR fetch of FLOAT/DOUBLE/REAL with a truncated buffer to post SQLSTATE 01004.SQL_C_WCHAR chunked SQLGetData buffer capacity to use sizeof(SQLWCHAR) so an 8-byte buffer fits three data characters plus NUL.SQL_C_WCHAR decimal-string binding to SQL_DECIMAL to convert correctly on all platforms.SQLFreeHandle / SQLDisconnect / SQLCopyDesc handle-hierarchy sequences that aborted or segfaulted.SQLDisconnect under iODBC: SQL_ATTR_CONNECTION_DEAD reports SQL_CD_TRUE and reads of other connection attributes return SQL_ERROR instead of stale cached values.SQLGetData conversion from DECFLOAT to SQL_C_WCHAR under iODBC to convert the value instead of returning SQL_SUCCESS with SQL_NULL_DATA and an untouched buffer.SQLColAttribute to map ODBC 2.x field identifiers (SQL_COLUMN_NAME, SQL_COLUMN_TYPE, etc.) to their SQL_DESC_* equivalents as previous behavior was inconsistent (empty strings, wrong values, or HY091).SQLPrimaryKeys / SQLForeignKeys with SQL_ATTR_METADATA_ID=TRUE to case-fold unquoted identifiers to uppercase before matching instead of comparing case-sensitively and returning empty results for lowercase names.SQLProcedures / SQLProcedureColumns with SQL_ATTR_METADATA_ID=TRUE to case-fold unquoted identifiers to uppercase before matching instead of returning empty results for lowercase names.SQLTables with SQL_ATTR_METADATA_ID=TRUE to case-fold unquoted identifiers to uppercase before matching instead of returning empty results for lowercase names.SQLGetData with SQL_C_NUMERIC to honor SQL_DESC_PRECISION and SQL_DESC_SCALE set on the ARD via SQLSetDescFieldinstead of hardcoded precision 38 / scale 0.DATE to SQL_C_CHAR/SQL_C_WCHAR conversion with an undersized buffer to return SQL_ERROR (22003) instead of truncating.SQLGetDescField on an empty IPD (no parameters bound) to return SQL_NO_DATA instead of SQL_ERROR.SQLGetStmtAttr / SQLGetConnectAttr with a negative string BufferLength to return SQL_ERROR (HY090).SQLGetConnectAttr with an out-of-range attribute identifier to return SQL_ERROR (HY092).SQLSetConnectAttr with ODBC 2.x statement-level attribute IDs (SQL_ATTR_MAX_ROWS, SQL_ATTR_QUERY_TIMEOUT) to return SQL_ERROR (HY092) instead of SQL_SUCCESS as a silent no-op.SQL_SUCCESS_WITH_INFO with SQLSTATE 01S07.FLOAT/REAL to single-field interval fetch with a nonzero fractional part to return SQL_SUCCESS_WITH_INFO (01S07).+0 (no negative-zero interval sign).Remedied SQL injection vulnerability in snowflake.connector.pandas_tools.write_pandas. See more https://github.com/snowflakedb/snowflake-connector-pyt…
v3.18.1(July 15,2026)
v3.18.0(October 03,2025)
v3.17.4(September 22,2025)
urllib3 to 2.5.0 and requests to v2.32.5v3.17.3(September 02,2025)
SF_SKIP_WARNING_FOR_READ_PERMISSIONS_ON_CONFIG_FILE environment variable.create_dataframe call on Snowparkv3.17.2(August 23,2025)
platform_detection_timeout_seconds to zero.v3.17.1(August 17,2025)
infer_schema parameter to write_pandas to perform schema inference on the passed data.snowlake reverted back to non-module.v3.17.0(August 16,2025)
unsafe_skip_file_permissions_check flag to skip file permission checks on the cache and configuration.APPLICATION_PATH within CLIENT_ENVIRONMENT to distinguish between multiple scripts using the Python Connector in the same environment.WORKLOAD_IDENTITY value for the authenticator type.workload_identity_provider and workload_identity_entra_resource parameters.use_vectorized_scanner parameter in the write_pandas function.type_code in ResultMetadata for interval types.snowflake_version property to the connection.OAUTH_TYPE to CLIENT_ENVIROMENT.pyarrow version constrain; versions >= 19 can now be used.external_session_id was not provided in SnowflakeRestful.fetch.Oauth and programmatic_access_token authenticator values.authenticator values.pandas.DataFrame or pyarrow.Table would overflow due to unnecessary precision. A clear error will now be raised if an overflow cannot be prevented.v3.16.0(July 04,2025)
bulk_upload_chunks parameter to write_pandas function. Setting this parameter to True changes the behaviour of write_pandas function to first write all the data chunks to the local disk and then perform the wildcard upload of the chunks folder to the stage. In default behaviour the chunks are being saved, uploaded and deleted one by one.client_fetch_use_mp parameter that enables multiprocessed fetching of result batches.write_pandas special characters usage in the location name.use_virtual_url when building the location for gcs storage client.v3.15.0(Apr 29,2025)
oauth_client_id, oauth_client_secret, oauth_authorization_url, oauth_token_request_url, oauth_redirect_uri, oauth_scope, oauth_disable_pkce, oauth_enable_refresh_tokens and oauth_enable_single_use_refresh_tokens parameters.OAUTH_AUTHORIZATION_CODE value for the parameter authenticator.oauth_client_id, oauth_client_secret, oauth_token_request_url, and oauth_scope parameters.OAUTH_CLIENT_CREDENTIALS value for the parameter authenticator.client_store_temporary_credential property is to be set to true.v3.14.1(April 21, 2025)
client_fetch_threads experimental parameter to better utilize threads for fetching query results.check_arrow_conversion_error_on_every_column connection property that can be set to False to restore previous behaviour in which driver will ignore errors until it occurs in the last column. This flag's purpose is to unblock workflows that may be impacted by the bugfix and will be removed in later releases.requestId), rather than automatically generate a UUID4 to use for the HTTP Request ID.TO_DATE arrow to python conversion when invalid date was followed by the correct one.insecure_mode connection property and replaced it with disable_ocsp_checks with the same behavior as the former property.v3.14.0(March 03, 2025)
write_pandas.unsafe_file_write connection parameter that restores the previous behaviour of saving files downloaded with GET with 644 permissions.v3.13.2(January 29, 2025)
v3.13.1(January 29, 2025)
v3.13.0(January 23,2025)
cosign.private_key authentication type.v3.12.4(December 3,2024)
v3.12.3(October 25,2024)
v3.12.2(September 11,2024)
snowflake.connector.util_text.random_string to reduce the likelihood of collisions.v3.12.1(August 20,2024)
login_timeout as a string raised a TypeError during the login retry step.pathlib instead of os for default config file location resolution.cryptogaphy version pin.snowflake-export-certs (its backing module was already removed long ago)v3.12.0(July 24,2024)
to_pandas() performance by fully parallel downloading logic.private_key.v3.11.0(June 17,2024)
token_file_path connection parameter to read an OAuth token from a file when connecting to Snowflake.debug_arrow_chunk connection parameter to allow debugging raw arrow data in case of arrow data parsing failure.disable_saml_url_check connection parameter to disable SAML URL check in OKTA authentication.: in a query parameter, for instance, https://sso.abc.com/idp/startSSO.ping?PartnerSpId=https://xyz.snowflakecomputing.com/.v3.10.1(May 21, 2024)
v3.10.0(April 29,2024)
v3.9.1(April 22,2024)
v3.9.0(April 20,2024)
$SNOWFLAKE_HOME/config.toml.v3.8.1(April 09, 2024)
write_pandas to skip TABLE IF NOT EXISTS in truncate mode." introduced in v3.8.0 (yanked) as it's a breaking change. write_pandas will be fixed in the future in a non-breaking way.v3.8.0(April 04,2024)
externalbrowser auth in containerized environments
/favicon on success pageSNOWFLAKE_AUTH_SOCKET_REUSE_PORT flag (usage: SNOWFLAKE_AUTH_SOCKET_REUSE_PORT=true) to set the underlying socket's SO_REUSEPORT flag (described in the socket man page)
SF_AUTH_SOCKET_PORT=3037 SNOWFLAKE_AUTH_SOCKET_REUSE_PORT=true poetry run python somescript.pySNOWFLAKE_AUTH_SOCKET_MSG_DONTWAIT flag (usage: SNOWFLAKE_AUTH_SOCKET_MSG_DONTWAIT=true) to make a non-blocking socket.recv call and retry on Error
write_pandas to skip TABLE IF NOT EXISTS in truncate mode.__del__ method.SnowflakeCursor.execute.v3.7.1(February 21, 2024)
write_pandas wasn't truncating the target table.v3.7.0(January 25,2024)
force_return_table to SnowflakeCursor.fetch_arrow_all to force returning pyarrow.Table in case of zero rows.v3.6.0(December 09,2023)
private_key_file and private_key_file_pwd connection parametersexpired to SnowflakeConnection class, that keeps track of whether the connection's master token has expired.v3.5.0(November 13,2023)
snowflake.connector.cursor.NanoarrowUsage.NANOARROW_USAGE.snowflake.connector.cursor.NANOARROW_USAGE.v3.4.1(November 08,2023)
urllib3 to 1.26.18requests to 2.31.0v3.4.0(November 03,2023)
use_logical_type in write_pandas.backoff_policy argument to snowflake.connector.connect allowing for configurable backoff policy between retries of failed requests. See available implementations in the backoff_policies module.socket_timeout argument to snowflake.connector.connect specifying socket read and connect timeout.login_timeout and network_timeout behaviour. Retries of login and network requests are now properly halted after these timeouts expire.urllib.v3.3.1(October 16,2023)
ImportError when the c extensions are not compiled.v3.3.0(October 10,2023)
NANOARROW_USAGE environment variable to allows switching between the nanoarrow converter and the arrow converter. Valid values include:
FOLLOW_SESSION_PARAMETER, which uses the converter configured in the server.DISABLE_NANOARROW, which uses arrow converter, overriding the server setting.ENABLE_NANOARROW, which uses the nanoarrow converter, overriding the server setting.snowflake.connector.cursor.NanoarrowUsage enum, whose members include:
NanoarrowUsage.FOLLOW_SESSION_PARAMETER, which uses the converter configured in the server.NanoarrowUsage.DISABLE_NANOARROW, which uses arrow converter, overriding the server setting.NanoarrowUsage.ENABLE_NANOARROW, which uses the nanoarrow converter, overriding the server setting.snowflake.connector.cursor.NANOARROW_USAGE module variable to allow switching between the nanoarrow converter and the arrow converter. It works in conjunction with the snowflake.connector.cursor.NanoarrowUsage enum.v3.2.1(September 26,2023)
v3.2.0(September 06,2023)
parser -> manager renaming more consistent in snowflake.connector.config_manager module.v3.1.1(August 28,2023)
RSAPublicKey when constructing AuthByKeyPair in addition to raw bytes.proxy_header is missing on SOCKSProxyManager.v3.1.0(July 31,2023)
Added a feature that lets you add connection definitions to the connections.toml configuration file. A connection definition refers to a collection of connection parameters, for example, if you wanted to define a connection named `prod``:
[prod]
account = "my_account"
user = "my_user"
password = "my_password"
By default, we look for the connections.toml file in the location specified in the SNOWFLAKE_HOME environment variable (default: ~/.snowflake). If this folder does not exist, the Python connector looks for the file in the platformdirs location, as follows:
~/.config/snowflake/, but follows XDG settings~/Library/Application Support/snowflake/%USERPROFILE%\AppData\Local\snowflake\You can determine which file is used by running the following command:
python -c "from snowflake.connector.constants import CONNECTIONS_FILE; print(str(CONNECTIONS_FILE))"
Bumped cryptography dependency from <41.0.0,>=3.1.0 to >=3.1.0,<42.0.0.
Improved OCSP response caching to remove tmp cache files on Windows.
Improved OCSP response caching to reduce the times of disk writing.
Added a parameter server_session_keep_alive in SnowflakeConnection that skips session deletion when client connection closes.
Tightened our pinning of platformdirs, to prevent their new releases breaking us.
Fixed a bug where SFPlatformDirs would incorrectly append application_name/version to its path.
Added retry reason for queries that are retried by the client.
Fixed a bug where write_pandas fails when user does not have the privilege to create stage or file format in the target schema, but has the right privilege for the current schema.
Remove Python 3.7 support.
Worked around a segfault which sometimes occurred during cache serialization in multi-threaded scenarios.
Improved error handling of connection reset error.
Fixed a bug about deleting the temporary files happened when running PUT command.
Allowed to pass type_mapper to fetch_pandas_batches() and fetch_pandas_all().
Fixed a bug where pickle.dump segfaults during cache serialization in multi-threaded scenarios.
Improved retry logic for okta authentication to refresh token if authentication gets throttled.
Note that this release does not include the changes introduced in the previous 3.1.0a1 release. Those will be released at a later time.
v3.0.4(May 23,2023)
cursor.execute() could modify the argument statement_params dictionary object when executing a multistatement query.SnowflakeCursor.nextset before fetching the result of the first query if the cursor runs an async multistatement query._prefetch_hook() was not called before yielding results of execute_async().v3.0.3(April 20, 2023)
snowflake.connector.pandas_tool.write_pandas.snowflake.connector.pandas_tool.write_pandas.snowflake.connector.pandas_tool.write_pandas.SnowflakeCursor.execute method.403: Forbidden HTTP response for a login-request.v3.0.2(March 23, 2023)
put command on AWS raised AttributeError when uploading file composed of multiple parts.SnowflakeCursor.fetch_arrow_all and SnowflakeCursor.fetchall.snowflake.connector.util_text.split_statements swallows the final line break in the case when there are no space between lines.v3.0.1(February 28, 2023)
AuthByKeyPair.handle_timeout should pass keyword arguments instead of positional arguments when calling AuthByKeyPair.prepare.v3.0.0(January 26, 2023)
fetch_arrow_all and fetch_arrow_batches to handle async query resultsv2.9.0(December 9, 2022)
cursor.execute and cursor.executemany
MULTI_STATEMENT_COUNT can be altered at the account, session, or statement level. An additional argument, num_statements, can be provided to execute to use this parameter at the statement level. It must be provided to executemany to submit a multi-statement query through the method. Note that bulk insert optimizations available through executemany are not available when submitting multi-statement queries.
get_results_from_sfqid to retrieve resultsSnowflakeCursor.nextset() as specified in the DB 2.0 API (PEP-249), to iterate through each statements results
execute (or get_results_from_sfqid if asynchronous) through the existing fetch*() methodsv2.8.3(November 28,2022)
v2.8.2(November 18,2022)
v2.8.1(October 30,2022)
v2.8.0(September 27,2022)
v2.7.12(August 26,2022)
v2.7.11(July 26,2022)
v2.7.10(July 22,2022)
lastrowid to SnowflakeCursor in compliance with PEP249.v2.7.9(June 26,2022)
v2.7.8(May 28,2022)
v2.7.7(April 30,2022)
v2.7.6(March 17,2022)
v2.7.5(March 17,2022)
v2.7.4(February 05,2022)
DatabaseError was thrown when executing against a closed cursor instead of InterfaceErrorexecutemany would crash if an iterator was supplied as argsNOT NULL constraint raised DatabaseError instead of IntegrityErrorv2.7.3(January 22,2022)
v2.7.2(December 17,2021)
v2.7.1(November 19,2021)
v2.7.0(October 25,2021)
v2.6.2(September 27,2021)
v2.6.1(September 16,2021)
v2.6.0(August 29,2021)
v2.5.1(July 31,2021)
v2.5.0(July 22,2021)
insert into ... commandsv2.4.6(June 25,2021)
v2.4.5(June 15,2021)
v2.4.4(May 30,2021)
v2.4.3(April 29,2021)
v2.4.2(April 03,2021)
v2.4.1(March 04,2021)
autocommitv2.4.0(March 04,2021)
v2.3.10(February 01,2021)
v2.3.9(January 27,2021)
v2.3.8(January 14,2021)
v2.3.7(December 10,2020)
v2.3.6(November 16,2020)
v2.3.5(November 03,2020)
v2.3.4(October 26,2020)
v2.3.3(October 05,2020)
v2.3.2(September 14,2020)
v2.3.1(August 25,2020)
v2.3.0(August 24,2020)
v2.2.10(August 03,2020)
v2.2.9(July 13,2020)
v2.2.8(June 22,2020)
v2.2.7(June 1,2020)
v2.2.6(May 11,2020)
v2.2.5(April 30,2020)
v2.2.4(April 10,2020)
v2.2.3(March 30,2020)
v2.2.2(March 9,2020)
v2.2.1(February 18,2020)
v2.2.0(January 27,2020)
v2.1.3(January 06,2020)
v2.1.2(December 16,2019)
v2.1.1(December 12,2019)
v2.1.0(December 2,2019)
ssl_context optionsv2.0.4(November 13,2019)
v2.0.3(November 1,2019)
v2.0.2(October 21,2019)
v2.0.1(October 04,2019)
v2.0.0(September 30,2019)
v1.9.1(October 4,2019)
v1.9.0(August 26,2019) REMOVED from pypi due to dependency compatibility issues
v1.8.7(August 12,2019)
v1.8.6(July 29,2019)
v1.8.5(July 15,2019)
v1.8.4(July 01,2019)
v1.8.3(June 17,2019)
v1.8.2 (June 03,2019)
v1.8.1 (May 20,2019)
v1.8.0 (May 10, 2019)
numpy.bool_ in binding typev1.7.11 (April 22, 2019)
v1.7.10 (April 8, 2019)
v1.7.9 (March 25,2019)
v1.7.8 (March 12,2019)
v1.7.7 (February 22,2019)
v1.7.6 (February 08,2019)
v1.7.5 (January 25, 2019)
v1.7.4 (January 3, 2019)
CLIENT_PREFETCH_THREADS to download the result set.v1.7.3 (December 11, 2018)
v1.7.2 (December 4, 2018)
v1.7.1 (November 27, 2018)
v1.7.0 (November 13, 2018)
v1.6.12 (October 30, 2018)
boto3 and botocore version dependeny.client_session_keep_alive_heartbeat_frequency to control heartbeat timings for client_session_keep_alive.v1.6.11 (October 23, 2018)
request_guid to each HTTP request for tracing.v1.6.10 (September 25, 2018)
client_session_keep_alive support.datetime for TIMESTAMP type in qmark binding mode. PR #118 (@rhlahuja)v1.6.9 (September 13, 2018)
v1.6.8 (August 30, 2018)
boto3 and botocore version dependeny.v1.6.7 (August 22, 2018)
v1.6.6 (August 9, 2018)
pycryptodome with pycryptodomex to avoid namespace conflict with PyCrypto.v1.6.5 (July 13, 2018)
v1.6.4 (July 5, 2018)
v1.6.3 (June 14, 2018)
v1.6.2 (June 7, 2018)
v1.6.1 (May 17, 2018)
login_timeout attribute is int.v1.6.0 (May 3, 2018)
v1.5.8 (April 26, 2018)
v1.5.7 (April 19, 2018)
v1.5.6 (April 5, 2018)
boto3 and botocore version dependeny.v1.5.5 (March 22, 2018)
cryptography dependency.v1.5.4 (March 15, 2018)
pyasn and pyasn1-modules version requirementspycryptodome version requirements. No 3.5.0 should be used.v1.5.3 (March 9, 2018)
pyasn1 for OCSP check in Python 2. Python 3 continue using asn1crypto for better performance.pycryptodome version to less than 3.5.0 for Issue 65.v1.5.2 (March 1, 2018)
boto3 and botocore version dependeny.v1.5.1 (February 15, 2018)
v1.5.0 (January 26, 2018)
pyasn1 and pyasn1-modules from the dependency.v1.4.17 (January 19, 2018)
pyasn1 and pyasn1-modules version dependency. PR 48 (@baxen)pyasn1 with asn1crypto Not activated yet.v1.4.16 (January 16, 2018)
v1.4.15 (January 11, 2018)
v1.4.14 (December 14, 2017)
v1.4.13 (November 30, 2017)
boto3 and botocore version dependeny.v1.4.12 (November 16, 2017)
qmark and numeric paramstyle support for server side binding.timezone session parameter support to connections.v1.4.11 (November 9, 2017)
v1.4.10 (October 26, 2017)
cryptography, boto3 and botocore version dependeny.v1.4.9 (October 10, 2017)
pyasn1 upgrade.v1.4.8 (October 5, 2017)
pyasn1. Relaxed the dependency.v1.4.7 (September 20, 2017)
v1.4.6 (September 14, 2017)
v1.4.5 (August 31, 2017)
v1.4.4 (August 24, 2017)
v1.4.3 (August 17, 2017)
certifi to the dependent component list to mitigate CA root certificate out of date issue.boto3 and botocore.cryptography and pyOpenSSL version dependeny change.validate_default_parameters to validate the default database, schema and warehouse. If the specified object doesn't exist, it raises an error.v1.4.2 (August 3, 2017)
pyasn1 and pyasn1-modulesv1.4.1 (July 26, 2017)
pyasn1 and pyasn1-modules versions to 0.2.3 and 0.0.9, respectivelyv1.4.0 (July 6, 2017)
boto3, botocore, cffi and cryptography and pyOpenSSLv1.3.18 (June 15, 2017)
v1.3.17 (June 1, 2017)
v1.3.16 (April 20, 2017)
DATE causing [Error 22] Invalid argument on WindowsRuntimeError in requestsv1.3.15 (March 30, 2017)
OpenSSL.SSL.SysCallError 10053 with lower concurrencyraw_msg attribute to Error classv1.3.14 (February 24, 2017)
sfqid and sqlstate to the results from query results.login_timeout and network_timeout.cryptography to 1.7.2, boto3 to 1.4.4 and botocore to 1.5.14.v1.3.13 (February 9, 2017)
login_timeout and network_timeout parameters to the Connection objects.v1.3.12 (February 2, 2017)
region parameter. One character was truncated from the tail of account namev1.3.11 (January 27, 2017)
v1.3.10 (January 26, 2017)
region option to support EU deployments betterv1.3.9 (January 16, 2017)
botocore to 1.4.93 to fix and boto3 to 1.4.3 to fix the HTTPS request failure in Python 3.6v1.3.8 (January 12, 2017)
six package to support both PY2 and PY3 for some functionscryptography to 1.7.1 to address MacOS Python 3.6 build issue.v1.3.7 (December 8, 2016)
v1.3.6 (December 1, 2016)
cryptography to 1.5.3, pyOpenSSL to 16.2.0 and cffi to 1.9.1.v1.3.5 (November 17, 2016)
Bad Request errorv1.3.4 (November 3, 2016)
quoted_name data type support for binding by SQLAlchemyparquiet file in PUT commandv1.3.3 (October 20, 2016)
botocore to 1.4.37 due to potential regression.v1.3.2 (October 12, 2016)
botocore to 1.4.52.PUT, GET commands and fetching large result set.v1.3.1 (September 30, 2016)
v1.3.0 (September 26, 2016)
Added support for the BINARY data type, which enables support for more Python data types:
Python 3:
bytes and bytearray can be used for binding.bytes is also used for fetching BINARY data type.Python 2:
bytearray can be used for bindingstr is used for fetching BINARY data type.Added proxy_user and proxy_password connection parameters for proxy servers that require authentication.
v1.2.8 (August 16, 2016)
botocore to 1.4.37.Connection.execute_string and Connection.execute_stream to run multiple statements in a string and stream.v1.2.7 (July 31, 2016)
snowflake.cursor.rowcount for INSERT ALL.use_accelerate_endpoint in PUT and GET if Transfer acceleration is enabled for the S3 bucket.python-future that loads test.py in the current directory.v1.2.6 (July 13, 2016)
v1.2.5 (July 8, 2016)
S3UploadFailedError and RetriesExceededError in PUT and GET, respectively.v1.2.4 (July 6, 2016)
max_connection_pool parameter to Connection so that you can specify the maximum number of HTTP/HTTPS connections in the pool.v1.2.3 (June 29, 2016)
v1.2.2 (June 21, 2016)
botocore to 1.4.26.v1.2.1 (June 13, 2016)
v1.2.0 (June 10, 2016)
v1.1.5 (June 2, 2016)
boto3 to 1.3.1 and botocore and 1.4.22.snowflake.cursor.rowcount for DML by snowflake.cursor.executemany.numpy data type binding support. numpy.intN, numpy.floatN and numpy.datetime64 can be bound and fetched.v1.1.4 (May 21, 2016)
cffi to 1.6.0.v1.1.3 (May 5, 2016)
cryptography to 1.3.2.v1.1.2 (May 4, 2016)
tzlocal optional.v1.1.1 (Apr 11, 2016)
cryptography to 1.3.1 and pyOpenSSL to 16.0.0.v1.1.0 (Apr 4, 2016)
bzip2 support in PUT command. This feature requires a server upgrade.snowflake._vendor with the dependency of boto3 1.3.0 and botocore 1.4.2.v1.0.7 (Mar 21, 2016)
pyOpenSSL at 0.15.1.v1.0.6 (Mar 15, 2016)
cryptography to 1.2.3.TIME data type, which is now a Snowflake supported data type. This feature requires a server upgrade.snowflake.connector.DistCursor to fetch the results in dict instead of tuple.v1.0.5 (Mar 1, 2016)
cryptography to 1.2.2 and cffi to 1.5.2.TIMESTAMP_LTZ to datetime in queries.v1.0.4 (Feb 15, 2016)
ETIMEDOUT and ECONNRESET.PUT command. They used to be in the same directory as the source data files.PUT commands if the token expires.v1.0.3 (Jan 13, 2016)
Added support for the BOOLEAN data type (i.e. TRUE or FALSE). This changes the behavior of the binding for the bool type object:
bool was bound as a numeric value (i.e. 1 for True, 0 for False).bool is bound as native SQL data (i.e. TRUE or FALSE).Added the autocommit method to the Connection object:
autocommit mode is ON (i.e. each DML statement commits the change).autocommit mode is OFF, the commit and rollback methods are enabled.Avoid segfault issue for cryptography 1.2 in Mac OSX by using 1.1 until resolved.
v1.0.2 (Dec 15, 2015)
boto3 1.2.2, botocore 1.3.12.SSLv3 mapping from the initial table.v1.0.1 (Dec 8, 2015)
v1.0.0 (Dec 1, 2015)
v3.18.1(July 15,2026)
v3.18.0(October 03,2025)
v3.17.4(September 22,2025)
urllib3 to 2.5.0 and requests to v2.32.5v3.17.3(September 02,2025)
SF_SKIP_WARNING_FOR_READ_PERMISSIONS_ON_CONFIG_FILE environment variable.create_dataframe call on Snowparkv3.17.2(August 23,2025)
platform_detection_timeout_seconds to zero.v3.17.1(August 17,2025)
infer_schema parameter to write_pandas to perform schema inference on the passed data.snowlake reverted back to non-module.v3.17.0(August 16,2025)
unsafe_skip_file_permissions_check flag to skip file permission checks on the cache and configuration.APPLICATION_PATH within CLIENT_ENVIRONMENT to distinguish between multiple scripts using the Python Connector in the same environment.WORKLOAD_IDENTITY value for the authenticator type.workload_identity_provider and workload_identity_entra_resource parameters.use_vectorized_scanner parameter in the write_pandas function.type_code in ResultMetadata for interval types.snowflake_version property to the connection.OAUTH_TYPE to CLIENT_ENVIROMENT.pyarrow version constrain; versions >= 19 can now be used.external_session_id was not provided in SnowflakeRestful.fetch.Oauth and programmatic_access_token authenticator values.authenticator values.pandas.DataFrame or pyarrow.Table would overflow due to unnecessary precision. A clear error will now be raised if an overflow cannot be prevented.v3.16.0(July 04,2025)
bulk_upload_chunks parameter to write_pandas function. Setting this parameter to True changes the behaviour of write_pandas function to first write all the data chunks to the local disk and then perform the wildcard upload of the chunks folder to the stage. In default behaviour the chunks are being saved, uploaded and deleted one by one.client_fetch_use_mp parameter that enables multiprocessed fetching of result batches.write_pandas special characters usage in the location name.use_virtual_url when building the location for gcs storage client.v3.15.0(Apr 29,2025)
oauth_client_id, oauth_client_secret, oauth_authorization_url, oauth_token_request_url, oauth_redirect_uri, oauth_scope, oauth_disable_pkce, oauth_enable_refresh_tokens and oauth_enable_single_use_refresh_tokens parameters.OAUTH_AUTHORIZATION_CODE value for the parameter authenticator.oauth_client_id, oauth_client_secret, oauth_token_request_url, and oauth_scope parameters.OAUTH_CLIENT_CREDENTIALS value for the parameter authenticator.client_store_temporary_credential property is to be set to true.v3.14.1(April 21, 2025)
client_fetch_threads experimental parameter to better utilize threads for fetching query results.check_arrow_conversion_error_on_every_column connection property that can be set to False to restore previous behaviour in which driver will ignore errors until it occurs in the last column. This flag's purpose is to unblock workflows that may be impacted by the bugfix and will be removed in later releases.requestId), rather than automatically generate a UUID4 to use for the HTTP Request ID.TO_DATE arrow to python conversion when invalid date was followed by the correct one.insecure_mode connection property and replaced it with disable_ocsp_checks with the same behavior as the former property.v3.14.0(March 03, 2025)
write_pandas.unsafe_file_write connection parameter that restores the previous behaviour of saving files downloaded with GET with 644 permissions.v3.13.2(January 29, 2025)
v3.13.1(January 29, 2025)
v3.13.0(January 23,2025)
cosign.private_key authentication type.v3.12.4(December 3,2024)
v3.12.3(October 25,2024)
v3.12.2(September 11,2024)
snowflake.connector.util_text.random_string to reduce the likelihood of collisions.v3.12.1(August 20,2024)
login_timeout as a string raised a TypeError during the login retry step.pathlib instead of os for default config file location resolution.cryptogaphy version pin.snowflake-export-certs (its backing module was already removed long ago)v3.12.0(July 24,2024)
to_pandas() performance by fully parallel downloading logic.private_key.v3.11.0(June 17,2024)
token_file_path connection parameter to read an OAuth token from a file when connecting to Snowflake.debug_arrow_chunk connection parameter to allow debugging raw arrow data in case of arrow data parsing failure.disable_saml_url_check connection parameter to disable SAML URL check in OKTA authentication.: in a query parameter, for instance, https://sso.abc.com/idp/startSSO.ping?PartnerSpId=https://xyz.snowflakecomputing.com/.v3.10.1(May 21, 2024)
v3.10.0(April 29,2024)
v3.9.1(April 22,2024)
v3.9.0(April 20,2024)
$SNOWFLAKE_HOME/config.toml.v3.8.1(April 09, 2024)
write_pandas to skip TABLE IF NOT EXISTS in truncate mode." introduced in v3.8.0 (yanked) as it's a breaking change. write_pandas will be fixed in the future in a non-breaking way.v3.8.0(April 04,2024)
externalbrowser auth in containerized environments
/favicon on success pageSNOWFLAKE_AUTH_SOCKET_REUSE_PORT flag (usage: SNOWFLAKE_AUTH_SOCKET_REUSE_PORT=true) to set the underlying socket's SO_REUSEPORT flag (described in the socket man page)
SF_AUTH_SOCKET_PORT=3037 SNOWFLAKE_AUTH_SOCKET_REUSE_PORT=true poetry run python somescript.pySNOWFLAKE_AUTH_SOCKET_MSG_DONTWAIT flag (usage: SNOWFLAKE_AUTH_SOCKET_MSG_DONTWAIT=true) to make a non-blocking socket.recv call and retry on Error
write_pandas to skip TABLE IF NOT EXISTS in truncate mode.__del__ method.SnowflakeCursor.execute.v3.7.1(February 21, 2024)
write_pandas wasn't truncating the target table.v3.7.0(January 25,2024)
force_return_table to SnowflakeCursor.fetch_arrow_all to force returning pyarrow.Table in case of zero rows.v3.6.0(December 09,2023)
private_key_file and private_key_file_pwd connection parametersexpired to SnowflakeConnection class, that keeps track of whether the connection's master token has expired.v3.5.0(November 13,2023)
snowflake.connector.cursor.NanoarrowUsage.NANOARROW_USAGE.snowflake.connector.cursor.NANOARROW_USAGE.v3.4.1(November 08,2023)
urllib3 to 1.26.18requests to 2.31.0v3.4.0(November 03,2023)
use_logical_type in write_pandas.backoff_policy argument to snowflake.connector.connect allowing for configurable backoff policy between retries of failed requests. See available implementations in the backoff_policies module.socket_timeout argument to snowflake.connector.connect specifying socket read and connect timeout.login_timeout and network_timeout behaviour. Retries of login and network requests are now properly halted after these timeouts expire.urllib.v3.3.1(October 16,2023)
ImportError when the c extensions are not compiled.v3.3.0(October 10,2023)
NANOARROW_USAGE environment variable to allows switching between the nanoarrow converter and the arrow converter. Valid values include:
FOLLOW_SESSION_PARAMETER, which uses the converter configured in the server.DISABLE_NANOARROW, which uses arrow converter, overriding the server setting.ENABLE_NANOARROW, which uses the nanoarrow converter, overriding the server setting.snowflake.connector.cursor.NanoarrowUsage enum, whose members include:
NanoarrowUsage.FOLLOW_SESSION_PARAMETER, which uses the converter configured in the server.NanoarrowUsage.DISABLE_NANOARROW, which uses arrow converter, overriding the server setting.NanoarrowUsage.ENABLE_NANOARROW, which uses the nanoarrow converter, overriding the server setting.snowflake.connector.cursor.NANOARROW_USAGE module variable to allow switching between the nanoarrow converter and the arrow converter. It works in conjunction with the snowflake.connector.cursor.NanoarrowUsage enum.v3.2.1(September 26,2023)
v3.2.0(September 06,2023)
parser -> manager renaming more consistent in snowflake.connector.config_manager module.v3.1.1(August 28,2023)
RSAPublicKey when constructing AuthByKeyPair in addition to raw bytes.proxy_header is missing on SOCKSProxyManager.v3.1.0(July 31,2023)
Added a feature that lets you add connection definitions to the connections.toml configuration file. A connection definition refers to a collection of connection parameters, for example, if you wanted to define a connection named `prod``:
[prod]
account = "my_account"
user = "my_user"
password = "my_password"By default, we look for the connections.toml file in the location specified in the SNOWFLAKE_HOME environment variable (default: ~/.snowflake). If this folder does not exist, the Python connector looks for the file in the platformdirs location, as follows:
~/.config/snowflake/, but follows XDG settings~/Library/Application Support/snowflake/%USERPROFILE%\AppData\Local\snowflake\You can determine which file is used by running the following command:
python -c "from snowflake.connector.constants import CONNECTIONS_FILE; print(str(CONNECTIONS_FILE))"
Bumped cryptography dependency from <41.0.0,>=3.1.0 to >=3.1.0,<42.0.0.
Improved OCSP response caching to remove tmp cache files on Windows.
Improved OCSP response caching to reduce the times of disk writing.
Added a parameter server_session_keep_alive in SnowflakeConnection that skips session deletion when client connection closes.
Tightened our pinning of platformdirs, to prevent their new releases breaking us.
Fixed a bug where SFPlatformDirs would incorrectly append application_name/version to its path.
Added retry reason for queries that are retried by the client.
Fixed a bug where write_pandas fails when user does not have the privilege to create stage or file format in the target schema, but has the right privilege for the current schema.
Remove Python 3.7 support.
Worked around a segfault which sometimes occurred during cache serialization in multi-threaded scenarios.
Improved error handling of connection reset error.
Fixed a bug about deleting the temporary files happened when running PUT command.
Allowed to pass type_mapper to fetch_pandas_batches() and fetch_pandas_all().
Fixed a bug where pickle.dump segfaults during cache serialization in multi-threaded scenarios.
Improved retry logic for okta authentication to refresh token if authentication gets throttled.
Note that this release does not include the changes introduced in the previous 3.1.0a1 release. Those will be released at a later time.
v3.0.4(May 23,2023)
cursor.execute() could modify the argument statement_params dictionary object when executing a multistatement query.SnowflakeCursor.nextset before fetching the result of the first query if the cursor runs an async multistatement query._prefetch_hook() was not called before yielding results of execute_async().v3.0.3(April 20, 2023)
snowflake.connector.pandas_tool.write_pandas.snowflake.connector.pandas_tool.write_pandas.snowflake.connector.pandas_tool.write_pandas.SnowflakeCursor.execute method.403: Forbidden HTTP response for a login-request.v3.0.2(March 23, 2023)
put command on AWS raised AttributeError when uploading file composed of multiple parts.SnowflakeCursor.fetch_arrow_all and SnowflakeCursor.fetchall.snowflake.connector.util_text.split_statements swallows the final line break in the case when there are no space between lines.v3.0.1(February 28, 2023)
AuthByKeyPair.handle_timeout should pass keyword arguments instead of positional arguments when calling AuthByKeyPair.prepare.v3.0.0(January 26, 2023)
fetch_arrow_all and fetch_arrow_batches to handle async query resultsv2.9.0(December 9, 2022)
cursor.execute and cursor.executemany
MULTI_STATEMENT_COUNT can be altered at the account, session, or statement level. An additional argument, num_statements, can be provided to execute to use this parameter at the statement level. It must be provided to executemany to submit a multi-statement query through the method. Note that bulk insert optimizations available through executemany are not available when submitting multi-statement queries.
get_results_from_sfqid to retrieve resultsSnowflakeCursor.nextset() as specified in the DB 2.0 API (PEP-249), to iterate through each statements results
execute (or get_results_from_sfqid if asynchronous) through the existing fetch*() methodsv2.8.3(November 28,2022)
v2.8.2(November 18,2022)
v2.8.1(October 30,2022)
v2.8.0(September 27,2022)
v2.7.12(August 26,2022)
v2.7.11(July 26,2022)
v2.7.10(July 22,2022)
lastrowid to SnowflakeCursor in compliance with PEP249.v2.7.9(June 26,2022)
v2.7.8(May 28,2022)
v2.7.7(April 30,2022)
v2.7.6(March 17,2022)
v2.7.5(March 17,2022)
v2.7.4(February 05,2022)
DatabaseError was thrown when executing against a closed cursor instead of InterfaceErrorexecutemany would crash if an iterator was supplied as argsNOT NULL constraint raised DatabaseError instead of IntegrityErrorv2.7.3(January 22,2022)
v2.7.2(December 17,2021)
v2.7.1(November 19,2021)
v2.7.0(October 25,2021)
v2.6.2(September 27,2021)
v2.6.1(September 16,2021)
v2.6.0(August 29,2021)
v2.5.1(July 31,2021)
v2.5.0(July 22,2021)
insert into ... commandsv2.4.6(June 25,2021)
v2.4.5(June 15,2021)
v2.4.4(May 30,2021)
v2.4.3(April 29,2021)
v2.4.2(April 03,2021)
v2.4.1(March 04,2021)
autocommitv2.4.0(March 04,2021)
v2.3.10(February 01,2021)
v2.3.9(January 27,2021)
v2.3.8(January 14,2021)
v2.3.7(December 10,2020)
v2.3.6(November 16,2020)
v2.3.5(November 03,2020)
v2.3.4(October 26,2020)
v2.3.3(October 05,2020)
v2.3.2(September 14,2020)
v2.3.1(August 25,2020)
v2.3.0(August 24,2020)
v2.2.10(August 03,2020)
v2.2.9(July 13,2020)
v2.2.8(June 22,2020)
v2.2.7(June 1,2020)
v2.2.6(May 11,2020)
v2.2.5(April 30,2020)
v2.2.4(April 10,2020)
v2.2.3(March 30,2020)
v2.2.2(March 9,2020)
v2.2.1(February 18,2020)
v2.2.0(January 27,2020)
v2.1.3(January 06,2020)
v2.1.2(December 16,2019)
v2.1.1(December 12,2019)
v2.1.0(December 2,2019)
ssl_context optionsv2.0.4(November 13,2019)
v2.0.3(November 1,2019)
v2.0.2(October 21,2019)
v2.0.1(October 04,2019)
v2.0.0(September 30,2019)
v1.9.1(October 4,2019)
v1.9.0(August 26,2019) REMOVED from pypi due to dependency compatibility issues
v1.8.7(August 12,2019)
v1.8.6(July 29,2019)
v1.8.5(July 15,2019)
v1.8.4(July 01,2019)
v1.8.3(June 17,2019)
v1.8.2 (June 03,2019)
v1.8.1 (May 20,2019)
v1.8.0 (May 10, 2019)
numpy.bool_ in binding typev1.7.11 (April 22, 2019)
v1.7.10 (April 8, 2019)
v1.7.9 (March 25,2019)
v1.7.8 (March 12,2019)
v1.7.7 (February 22,2019)
v1.7.6 (February 08,2019)
v1.7.5 (January 25, 2019)
v1.7.4 (January 3, 2019)
CLIENT_PREFETCH_THREADS to download the result set.v1.7.3 (December 11, 2018)
v1.7.2 (December 4, 2018)
v1.7.1 (November 27, 2018)
v1.7.0 (November 13, 2018)
v1.6.12 (October 30, 2018)
boto3 and botocore version dependeny.client_session_keep_alive_heartbeat_frequency to control heartbeat timings for client_session_keep_alive.v1.6.11 (October 23, 2018)
request_guid to each HTTP request for tracing.v1.6.10 (September 25, 2018)
client_session_keep_alive support.datetime for TIMESTAMP type in qmark binding mode. PR #118 (@rhlahuja)v1.6.9 (September 13, 2018)
v1.6.8 (August 30, 2018)
boto3 and botocore version dependeny.v1.6.7 (August 22, 2018)
v1.6.6 (August 9, 2018)
pycryptodome with pycryptodomex to avoid namespace conflict with PyCrypto.v1.6.5 (July 13, 2018)
v1.6.4 (July 5, 2018)
v1.6.3 (June 14, 2018)
v1.6.2 (June 7, 2018)
v1.6.1 (May 17, 2018)
login_timeout attribute is int.v1.6.0 (May 3, 2018)
v1.5.8 (April 26, 2018)
v1.5.7 (April 19, 2018)
v1.5.6 (April 5, 2018)
boto3 and botocore version dependeny.v1.5.5 (March 22, 2018)
cryptography dependency.v1.5.4 (March 15, 2018)
pyasn and pyasn1-modules version requirementspycryptodome version requirements. No 3.5.0 should be used.v1.5.3 (March 9, 2018)
pyasn1 for OCSP check in Python 2. Python 3 continue using asn1crypto for better performance.pycryptodome version to less than 3.5.0 for Issue 65.v1.5.2 (March 1, 2018)
boto3 and botocore version dependeny.v1.5.1 (February 15, 2018)
v1.5.0 (January 26, 2018)
pyasn1 and pyasn1-modules from the dependency.v1.4.17 (January 19, 2018)
pyasn1 and pyasn1-modules version dependency. PR 48 (@baxen)pyasn1 with asn1crypto Not activated yet.v1.4.16 (January 16, 2018)
v1.4.15 (January 11, 2018)
v1.4.14 (December 14, 2017)
v1.4.13 (November 30, 2017)
boto3 and botocore version dependeny.v1.4.12 (November 16, 2017)
qmark and numeric paramstyle support for server side binding.timezone session parameter support to connections.v1.4.11 (November 9, 2017)
v1.4.10 (October 26, 2017)
cryptography, boto3 and botocore version dependeny.v1.4.9 (October 10, 2017)
pyasn1 upgrade.v1.4.8 (October 5, 2017)
pyasn1. Relaxed the dependency.v1.4.7 (September 20, 2017)
v1.4.6 (September 14, 2017)
v1.4.5 (August 31, 2017)
v1.4.4 (August 24, 2017)
v1.4.3 (August 17, 2017)
certifi to the dependent component list to mitigate CA root certificate out of date issue.boto3 and botocore.cryptography and pyOpenSSL version dependeny change.validate_default_parameters to validate the default database, schema and warehouse. If the specified object doesn't exist, it raises an error.v1.4.2 (August 3, 2017)
pyasn1 and pyasn1-modulesv1.4.1 (July 26, 2017)
pyasn1 and pyasn1-modules versions to 0.2.3 and 0.0.9, respectivelyv1.4.0 (July 6, 2017)
boto3, botocore, cffi and cryptography and pyOpenSSLv1.3.18 (June 15, 2017)
v1.3.17 (June 1, 2017)
v1.3.16 (April 20, 2017)
DATE causing [Error 22] Invalid argument on WindowsRuntimeError in requestsv1.3.15 (March 30, 2017)
OpenSSL.SSL.SysCallError 10053 with lower concurrencyraw_msg attribute to Error classv1.3.14 (February 24, 2017)
sfqid and sqlstate to the results from query results.login_timeout and network_timeout.cryptography to 1.7.2, boto3 to 1.4.4 and botocore to 1.5.14.v1.3.13 (February 9, 2017)
login_timeout and network_timeout parameters to the Connection objects.v1.3.12 (February 2, 2017)
region parameter. One character was truncated from the tail of account namev1.3.11 (January 27, 2017)
v1.3.10 (January 26, 2017)
region option to support EU deployments better</Note truncated.
Added support for pandas conversion for Day-time and Year-Month Interval types
Added support for intermediate certificates as roots when they are stored in the trust store
urllib3 to 2.5.0 and requests to v2.32.5Enhanced configuration file permission warning messages.
SF_SKIP_WARNING_FOR_READ_PERMISSIONS_ON_CONFIG_FILE environment variable.create_dataframe call on SnowparkFixed a bug where platform_detection was retrying failed requests with warnings to non-existent endpoints.
platform_detection_timeout_seconds to zero.Added infer_schema parameter to write_pandas to perform schema inference on the passed data.
infer_schema parameter to write_pandas to perform schema inference on the passed data.snowlake reverted back to non-module.Added in-band HTTP exception telemetry.
unsafe_skip_file_permissions_check flag to skip file permission checks on the cache and configuration.APPLICATION_PATH within CLIENT_ENVIRONMENT to distinguish between multiple scripts using the Python Connector in the same environment.WORKLOAD_IDENTITY value for the authenticator type.workload_identity_provider and workload_identity_entra_resource parameters.use_vectorized_scanner parameter in the write_pandas function.type_code in ResultMetadata for interval types.snowflake_version property to the connection.OAUTH_TYPE to CLIENT_ENVIROMENT.pyarrow version constrain; versions >= 19 can now be used.external_session_id was not provided in SnowflakeRestful.fetch.Oauth and programmatic_access_token authenticator values.authenticator values.pandas.DataFrame or pyarrow.Table would overflow due to unnecessary precision. A clear error will now be raised if an overflow cannot be prevented.Bumped numpy dependency from <2.1.0 to <=2.2.4.
bulk_upload_chunks parameter to write_pandas function. Setting this parameter to True changes the behaviour of write_pandas function to first write all the data chunks to the local disk and then perform the wildcard upload of the chunks folder to the stage. In default behaviour the chunks are being saved, uploaded and deleted one by one.client_fetch_use_mp parameter that enables multiprocessed fetching of result batches.write_pandas special characters usage in the location name.use_virtual_url when building the location for gcs storage client.Bumped up min boto and botocore version to 1.24.
oauth_client_id, oauth_client_secret, oauth_authorization_url, oauth_token_request_url, oauth_redirect_uri, oauth_scope, oauth_disable_pkce, oauth_enable_refresh_tokens and oauth_enable_single_use_refresh_tokens parameters.OAUTH_AUTHORIZATION_CODE value for the parameter authenticator.oauth_client_id, oauth_client_secret, oauth_token_request_url, and oauth_scope parameters.OAUTH_CLIENT_CREDENTIALS value for the parameter authenticator.client_store_temporary_credential property is to be set to true.Deprecated insecure_mode connection property and replaced it with disable_ocsp_checks with the same behavior as the former property.
client_fetch_threads experimental parameter to better utilize threads for fetching query results.check_arrow_conversion_error_on_every_column connection property that can be set to False to restore previous behaviour in which driver will ignore errors until it occurs in the last column. This flag's purpose is to unblock workflows that may be impacted by the bugfix and will be removed in later releases.requestId), rather than automatically generate a UUID4 to use for the HTTP Request ID.TO_DATE arrow to python conversion when invalid date was followed by the correct one.insecure_mode connection property and replaced it with disable_ocsp_checks with the same behavior as the former property.Bumped pyOpenSSL dependency upper boundary from <25.0.0 to <26.0.0.
write_pandas.unsafe_file_write connection parameter that restores the previous behaviour of saving files downloaded with GET with 644 permissions.Changed not to use scoped temporary objects.
Remedied SQL injection vulnerability in snowflake.connector.pandas_tools.write_pandas. See more https://github.com/snowflakedb/snowflake-connector-pyt…
Added a feature to limit the sizes of IO-bound ThreadPoolExecutors during PUT and GET commands.
cosign.private_key authentication type.Fixed a bug where multipart uploads to Azure would be missing their MD5 hashes.
Improved the error message for SSL-related issues to provide clearer guidance when an SSL error occurs.
Improved error handling for asynchronous queries, providing more detailed and informative error messages when an async query fails.
snowflake.connector.util_text.random_string to reduce the likelihood of collisions.Fixed a bug that logged the session token when renewing a session.
login_timeout as a string raised a TypeError during the login retry step.pathlib instead of os for default config file location resolution.cryptogaphy version pin.snowflake-export-certs (its backing module was already removed long ago)Set default connection timeout of 10 seconds and socket read timeout of 10 minutes for HTTP calls in file transfer.
to_pandas() performance by fully parallel downloading logic.private_key.Added support for token_file_path connection parameter to read an OAuth token from a file when connecting to Snowflake.
v3.11.0(June 17,2024)
token_file_path connection parameter to read an OAuth token from a file when connecting to Snowflake.debug_arrow_chunk connection parameter to allow debugging raw arrow data in case of arrow data parsing failure.disable_saml_url_check connection parameter to disable SAML URL check in OKTA authentication.: in a query parameter, for instance, https://sso.abc.com/idp/startSSO.ping?PartnerSpId=https://xyz.snowflakecomputing.com/.Removed an incorrect error log message that could occur during arrow data conversion.
v3.10.1(May 21, 2024)
Added support for structured types to fetch_pandas_all.
v3.10.0(April 29,2024)
Fixed an issue that caused a HTTP 400 error when connecting to a China endpoint.
v3.9.1(April 22,2024)
Added easy logging configuration so that users can easily generate log file by setup log config in $SNOWFLAKE_HOME/config.toml.
v3.9.0(April 20,2024)
$SNOWFLAKE_HOME/config.toml.…mode." introduced in v3.8.0 (yanked) as it's a breaking change. write_pandas will be fixed in the future in a non-breaking way.
v3.8.1(April 09, 2024)
write_pandas to skip TABLE IF NOT EXISTS in truncate mode." introduced in v3.8.0 (yanked) as it's a breaking change. write_pandas will be fixed in the future in a non-breaking way.v3.8.0(April 04,2024)
externalbrowser auth in containerized environments
/favicon on success pageSNOWFLAKE_AUTH_SOCKET_REUSE_PORT flag (usage: SNOWFLAKE_AUTH_SOCKET_REUSE_PORT=true) to set the underlying socket's SO_REUSEPORT flag (described in the socket man page)
SF_AUTH_SOCKET_PORT=3037 SNOWFLAKE_AUTH_SOCKET_REUSE_PORT=true poetry run python somescript.pySNOWFLAKE_AUTH_SOCKET_MSG_DONTWAIT flag (usage: SNOWFLAKE_AUTH_SOCKET_MSG_DONTWAIT=true) to make a non-blocking socket.recv call and retry on Error
write_pandas to skip TABLE IF NOT EXISTS in truncate mode.__del__ method.SnowflakeCursor.execute.Nothing published for this version
Bumped keyring dependency lower bound to 23.1.0 to address security vulnerability.
v3.7.1(February 21, 2024)
write_pandas wasn't truncating the target table.Added a new boolean parameter force_return_table to SnowflakeCursor.fetch_arrow_all to force returning pyarrow.Table in case of zero rows.
v3.7.0(January 25,2024)
force_return_table to SnowflakeCursor.fetch_arrow_all to force returning pyarrow.Table in case of zero rows.Changed urllib3 version pin to only affect Python versions < 3.10.
v3.6.0(December 09,2023)
private_key_file and private_key_file_pwd connection parametersexpired to SnowflakeConnection class, that keeps track of whether the connection's master token has expired.Deprecated the usage of the following class/variable/environment variable for the sake of pure nanoarrow converter:
v3.5.0(November 13,2023)
snowflake.connector.cursor.NanoarrowUsage.NANOARROW_USAGE.snowflake.connector.cursor.NANOARROW_USAGE.One column per quarter.