This document outlines features and improvements from each release.
Note
All releases before v1.0.0 are considered pre-release and are for non-production testing and evaluation, and may include changes to the API.
- Add multi-backendchat factory (SingleStoreChatFactory) initilizing various flavors of langchain chat clients.
- Enhance headers injection (Authorization and Impersonating JWT) across Azure and Amazon Bedrock models.
- Add / update release utilities
- Improve UDF error messages
- Add IN GROUP clause to USE WORKSPACE Fusion SQL command (#93)
- Remove extra / at end of path in Stages.listdir
- Improve errors in UDF app (#92)
- Fix precision values for decimals (#91)
- Fix error and memory handling in the C extension
- Fixed object attribute 'servers' error in UDF application (#87) (experimental)
- Added enforcement for function definition limits (#86)
- Improved error handling for mismatched numpy versions
- Updated documentation with scheduled jobs information
- Improved UDF application logging (experimental)
- Add more validation of UDF type parameters in UDF app (experimental)
- Add docstring information to function information returned by UDF app (experimental)
- Update OpenAI wrapper
- Fix starter workspace connections
- Fix management API HTTP requests with extra parameters
- Fix issues with SQL notebook cells in notebooks called from %run_shared or %run_personal
- Add starter workspace objects to Management API
- Add async support, timeouts, logging, and the ability to cancel UDFs in the UDF ASGI application (experimental)
- Fixes for Iceberg export Fusion SQL (experimental)
- Add additional Fusion SQL commands for Iceberg export (experimental)
- Fix unsigned ints in Data API
- Add vector_db interface
- Refactor external function server (experimental) type system
- Back out charset setting code
- Fix default charset issue
- Fix workspace group lookup in some Fusion SQL commands
- Fix struct formatting issue in vector result support
- Add cloud notebook information to client version
- Fix
IN GROUPclause in Fusion SQL for files - Fix file-like objects in file upload APIs
- Add support for external TVFs
- Add support for pyarrow Table and pandas / polars DataFrames return values in external functions
- Add support for return value field names
- Add support for dataclasses and Pydantic external function parameter and return value definitions
- Add
singlestoredb.server.dockerandsinglestoredb.server.free_tierserver interfaces - Fix vector result conversions when numpy is not installed
- Fix connection memory leak
- Add Fusion SQL commands for managing models
- Add
%run_sharedand%run_personalmagics - Update doc for Export service Fusion SQL commands
- Add Fusion SQL commands for personal and shared files
- Add Fusion SQL commands for Iceberg egress service
- Add
tls_sni_servernameto connection - Fix special character parsing in connection URLs
- Add Fusion SQL commands for scheduled notebooks
- Fixes for apps
- Fixes for apps
- Add dashboard and code as a service functions
- Fix HTTP (Data API) driver timeout
- Add
client_found_rowsoption to connection
- Fix string length calculation
- Fix memory leak in connections
- Add job service Fusion commands
- Add
USE WORKSPACEFusion command - Add
vector_data_format=to connection options to specify JSON or BINARY for vector data
- Purge expired JWTs from
get_tokenas needed
- Fix Javascript function name in portal object
- Fix binary vector conversions
- Add async events for JWT updates
- Add ai module
- Fix
encoding_errors=option
- Add ability to use iterables and queues as the source of
LOCAL INFILE
- Added support for vector data type conversions in query results
- Added
singlestoredb.notebookmodule for interacting with portal objects
- Added new result types: numpy, pandas, polars, arrow
- Added
SHOW FUSION HELPcommand and doc to Fusion SQL handlers
- Management API updates
- Updates to external function application API
- Add configuration parameter for Management API base URL
- Add
singlestoredb.management.get_secretfunction - Add
add_endpoint_to_firewall_ranges=parameter tocreate_workspace
- Fix issues in Fusion SQL Stage commands
- Add
callfunction to UDF app
- Update Stage API
- Fixes for deferred connections
- Add secrets managemnt to Management API
- Refactor UDF data converters
- Add Fusion SQL handlers for suspend and resume
- Various Fusion SQL handler fixes
- Add pytest plugin
- Fix deferred connection URL in HTTP
- Change Stage URLs
- Fix password encoding in deferred connections
- Make workspace group optional in Fusion SQL if it is specified in the environment
- Cache regions in Management API
- Add dummy fields to Region objects if the region ID does not point to an existing region
- Add JWT authentication support to Fusion
- Add experimental vector data format support to UDF server
- Rename
stagestostage - Add
track_envconnection parameter to automatically track theSINGLESTOREDB_URLenvironment variable
- Fusion fixes and testing
- Add defaults for builtin Fusion rules
- More Fusion enhancements
- Fusion fixes
- Experimental Fusion SQL interface
- Add name / ID indexing to workspace groups / workspaces / regions
- Add Stage to Management API
- Add debug option for connections
- Fix error propagation issue in C extension
- Add encoding_errors= parameter to connection
- Fix
WITHstatements in HTTP
- Fix
DESCRIBEstatements in HTTP
- Fix boolean connection options
- Fix
%escaping in HTTP queries
- Add
nan_as_nullandinf_as_nulloptions for parameter conversion support - Separate
structsequencesandnamedtuplesforresults_type - Performance improvements of binary data uploads
- Add
create_enginefunction to return SQLAlchemy engine while supporting environment variable parameter settings and settings in options
- ! Python 3.8 is now the minimum required version
- Add parameter conversion routines to HTTP driver
- Add
connect_timeoutandmulti_statementsoptions to connection
- Add converters for numpy array to vector blobs, and pygeos / shapely objects to geography data
- Fix GSSAPI/Kerberos packet data
- Added GSSAPI/Kerberos support
- Added expiration to workspaces
- Fixed issue with parsing numeric results
- Fixed issues with unbuffered reads
- Added 32-bit Windows and aarch64 Linux packages
- Added option to log queries
- ! Query parameter syntax has changed from
:1for positional and:keyfor dictionary keys to%sfor positional and%(key)sfor dictionary keys - !
results_formatconnection parameter has changed toresults_type - High-performance C extension added
- Added
ssl_verify_certandssl_verify_identityconnection options - Add Python 3.11 support
- Add Python 3.6 support
- Add
ssl_cipheroption to connections - Add
showaccessor for databaseSHOWcommands
- Fixes for PyMySQL compatibility
- Changed cipher in PyMySQL connection for SingleStoreDB Cloud compatibility
- Changed autocommit=True by default
- Changed to pure Python driver
- Add workspace management objects
- Added
auth.get_jwtfunction for retrieving JWTs
- DB-API compliant connections
- HTTP API support
- Cluster manager interface