pyspark.ml.connectThe public machine learning APIs exposed by pyspark.ml.connect have been deprecated since Spark 4.0.0. Use the standard pyspark.ml APIs, which support Spark Connect.
New Spark Connect clients should always exclude these deprecated public APIs from implementation and API-parity targets. Missing support for them is not a parity gap. Follow the standard APIs, such as pyspark.ml.classification.LogisticRegression, pyspark.ml.Pipeline, and pyspark.ml.tuning.CrossValidator.
This directory also contains internal helpers used by standard pyspark.ml over Spark Connect, including proto.py, serialize.py, and readwrite.py. These helpers remain relevant when implementing standard ML support in new clients.