1.30.21

1.30.21.2

Improvements

  • pandas version 1.0.0 or higher is now required as a dependency.

Bug Fixes

  • Fixed a segmentation error when calling the getUnwrittenData method after the MultithreadedTableWriter failed to insert data.

  • Downloading of BLOB data larger than 64 KB is now supported.

  • Fixed an issue where out-of-bounds access occurred when subscribing to data from DolphinDB server version 1.30.21/2.00.9 or later on MAC ARM.

  • Fixed incorrect data type conversion when uploading null values of np.datetime64 type.

  • Fixed decimal overflow when uploading a vector with the first element being a Decimal("NaN").

  • Fixed a segmentation error when downloading BLOB sets using the PROTOCOL_DDB protocol.

  • Fixed an issues where a session variable named “db“ was overwritten when calling the loadTableBySQL method.

  • Fixed an issue where the process would be stuck when data was not retrieved after calling the addTask method of DBConnectionPool.

1.30.21.1

New Features

  • Added support for Python 3.10.

  • Added a new parameter protocol to Session and DBConnectionPool constructors to specify the data transfer protocol.

  • Subscribed data can now be pushed using the connection initiated by the subscriber through Python API.

  • Added a new parameter args to pass user-defined objects to DBConnectionPool.addTask.

  • tableAppender, tableUpsert and PartitionedTableAppender now support uploading IPaddr, UUID, and INT128 values.

  • Added support for downloading data using the Apache Arrow format.

  • Added support for downloading and uploading DECIMAL values using the DolphinDB-customized data communication protocol.

  • Error message enhancements

Bug Fixes

  • Fixed semaphore creation error which is raised after multiple creations of MultithreadedTableWriter in macOS.

  • Fixed the "unmarshall failed" error when downloading an empty table containing STRING columns with pickle enabled.

  • Fixed the issue where the request is aborted when the subscribed data contains array vectors.

  • Fixed the issue in uWSGI when executing a SQL query with the Python API.

  • Fixed the issue when uploading np.nan values, the server displays "NaN" instead of converting them to NULL values.