3.0.0.1

Improvements

• Enhanced the memory management when using MultithreadedTableWriter.

3.0.0.0

New Features

• Added classes Event/Scalar/Vector to support custom events.
• Added class EventSender to write events into a heterogeneous stream table.
• Added class EventClient to subscribe events from a heterogeneous stream table.

Improvements

• Added parameters backupSites, resubTimeout, subOnce to enable failover mechanism for subscribe.
• Added support for the following pandas ExtensionDtype: Boolean/Int8/Int16/Int32/Int64/Float32/Float64/String.
• Enhanced type hints.

Bug Fixes

• Fixed the deserialization issue when unpacking data with the PROTOCOL_PICKLE.
• Fixed memory leaks when subscribing to the data of the following types: STRING, SYMBOL, IP, UUID, INT128, BLOB.
• Fixed an issue where time data was being written incorrectly with TableAppender due to pandas version incompatibility under certain circumstances.
• Fixed the issue where NaN values were not correctly converted to NULL values during the uploading process.
• Fixed a segmentation fault when writing strings to a DECIMAL column with TableAppender/TableUpserter/PartitionedTableAppender.

2.0.11.0

Note: The version number of DolphinDB Python API has been adjusted to improve user experience. The new version number aligns with DolphinDB Server 200.

Improvements

• Removed the length limit on the uploaded BLOB data; added the length limit on the uploaded SYMBOL/STRING data to less than 256 KB.
• Enhanced type annotations.
• Added new parameter sortColumns to Session.loadTextEx to support loading data to the TSDB database.
• Enabled Python API’s compatibility with NumPy version up to 1.26.3 and pandas version up to 2.2.0.
• Optimized the handling of illegal column names in the Table class.

Bug Fixes

• Fixed the lag when uploading empty DataFrame with a SYMBOL column.
• Fixed the issue where the BLOB column of the subscribed heterogeneous stream table contains NULL values.
• Fixed the issue where the decimal points in DECIMAL data of the subscribed heterogeneous stream table are missing.

1.30.22.6

New Features

• The DolphinDB Python API now supports Python 3.11.
• The PROTOCOL_DDB protocol now supports the Decimal128 data type.

Enhancements

• The drop method now behaves differently when the input is a string or single-element list.
• To avoid potential issues, strings exceeding 64KB cannot be uploaded.

Bug Fixes

• Fixed an issue related to uploading dictionaries that contained data types such as UUID, IPADDR, and INT128.
• Fixed improper destruction of members in the MultithreadedTableWriter class.
• Fixed the MultithreadedTableWriter aborting on invalid data inserts in some cases.
• Fixed a segmentation fault when uploading DataFrames, particularly when the first value in a Decimal32/64/128 column was NULL.
• Fixed an issue which could led to errors or data inconsistencies when uploading DataFrames with a datetime64[us] column.

1.30.22.5

Bug Fixes

• Fixed an issue where NumPy ndarrays configured with an unexpected memory order became disordered after upload.
• Fixed compatibility issues with TableAppender, TableUpserter, and PartitionedTableAppender when appending data to tables in a lower DolphinDB server version.
• Fixed an issue where, under specific circumstances, null values in array vectors were missing after upload.

1.30.22.4

New Features

• The PROTOCOL_ARROW serialization protocol is now supported on all operating systems for the DolphinDB Python API.
• Added support for uploading Pandas 2.0 DataFrames that use PyArrow as the storage backend.
• Explicit type conversions to Decimal32 and Decimal64 now support specifying scale.