GH-44922: [MATLAB] Add IPC RecordBatchStreamFileWriter MATLAB class

[sgilmore10]

Rationale for this change

To enable support for the IPC Streaming format in the MATLAB interface, we should add a RecordBatchStreamWriter class.

What changes are included in this PR?

Added arrow.io.ipc.RecordBatchStreamWriter class.

Example Usage:

>> city = ["Boston" "Seattle" "Denver" "Juno" "Anchorage" "Chicago"]';
>> daylength = duration(["15:17:01" "15:59:16" "14:59:14" "19:21:23" "14:18:24" "15:13:39"])';
>> matlabTable = table(city, daylength, VariableNames=["City", "DayLength"]);
>> recordBatch1 = arrow.recordBatch(matlabTable(1:4, :))
>> recordBatch2 = arrow.recordBatch(matlabTable(5:end, :));

>> writer = arrow.io.ipc.RecordBatchStreamWriter("daylight.arrow", recordBatch1.Schema);
>> writer.writeRecordBatch(recordBatch1);
>> writer.writeRecordBatch(recordBatch2);
>> writer.close();

Are these changes tested?

Yes. I Parameterized the test cases in test/arrow/io/ipc/tRecordBatchWriter.m to test the behavior of both arrow.io.ipc.RecordBatchFileWriter AND arrow.io.ipc.RecordBatchStreamWriter.

Are there any user-facing changes?

Yes. Users can now use arrow.io.ipc.RecordBatchStreamWriter to serialize RecordBatches/Tables to the Arrow IPC Streaming format.

Future Directions

1. [MATLAB] Add IPC RecordBatchStreamReader MATLAB class #44923

• GitHub Issue: [MATLAB] Add IPC RecordBatchStreamFileWriter MATLAB class #44922

[kevingurney]

This looks really good! Thanks, @sgilmore10!

[kevingurney]

+1

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 2b5a802.

There were 132 benchmark results with an error:

• Commit Run on arm64-t4g-2xlarge-linux at 2024-12-04 17:36:18Z
  • tpch (R) with engine=arrow, format=parquet, language=R, memory_map=False, query_id=TPCH-07, scale_factor=1
  • tpch (R) with engine=arrow, format=native, language=R, memory_map=False, query_id=TPCH-09, scale_factor=1
• and 130 more (see the report linked below)

There were 2 benchmark results indicating a performance regression:

• Commit Run on amd64-m5-4xlarge-linux at 2024-12-04 18:31:06Z
  • dataset-read (Python) with async=True, dataset=nyctaxi_multi_parquet_s3, pre_buffer=true
  • dataset-read (Python) with async=True, dataset=nyctaxi_multi_ipc_s3

The full Conbench report has more details. It also includes information about 29 possible false positives for unstable benchmarks that are known to sometimes produce them.