Hi Spark devs, We're extending Spark's datetime support on two fronts, and both have a groomed backlog of sub-tasks ready to pick up - great for a first Spark patch or a larger feature:
- Nanosecond timestamps TIMESTAMP_NTZ(p) / TIMESTAMP_LTZ(p), p in [7,9] - https://issues.apache.org/jira/browse/SPARK-56822 - The TIME data type TIME(p), p in [0,9] - https://issues.apache.org/jira/browse/SPARK-57550 The goal is parity with DATE/TIMESTAMP and closer ANSI alignment. Most remaining tasks are self-contained and follow an existing pattern. Good first tasks (labeled starter): SPARK-57809, SPARK-57810, SPARK-57807, SPARK-57808 (nanos) and SPARK-57846, SPARK-57847, SPARK-57850 (TIME). Find them all with: project = SPARK AND (parent = SPARK-56822 OR parent = SPARK-57550) AND labels = starter AND resolution = Unresolved Meatier work: vectorized Parquet read (SPARK-57828), streaming watermark/windows (SPARK-57830/57829/57843), CBO estimation (SPARK-57839); TIME sequence() (SPARK-57852) and the ANSI modulo-24 semantics question (SPARK-57853). Each umbrella lists the rest with dependency notes. To jump in: comment on a sub-task to get it assigned, then open a PR referencing the JIRA (https://spark.apache.org/contributing.html). Happy to help you get unblocked — just reply or comment on the ticket. Thanks, Max
