v0.3.0

Release Version 0.3.0
[0.3.0] - 2022-08-17
Added

• The data transfer module now supports any combination of H2 and Apache Spark
  as client and server. Apache Spark uses distributed Receivers run by Workers.
• The configuration file parser now also supports inline comments and emits
  errors that contain information about which settings are missing arguments.
• Improved messages from predicate push-down when a predicate is not supported.
• External tables can be added to Apache Spark with modelardb.spark.external.

Changed

• The Apache Arrow Flight-based query interface now transfers data in batches.
• Reverted to the key-value configuration file format and removed PureConfig.
  The data transfer module is configured with the modelardb.transfer setting.
• The value of modelardb.dynamic_split_fraction is now used directly as set.
• Renamed gaps in the Segment View to offsets for both Apache Spark and H2.
• Renamed mid to mtid in ORCStorage and ParquetStorage for consistency.

Removed

• Dropped Akka Streams from the data transfer module and switched to using
  Apache Arrow Flight directly for data transfer.

Fixed

• H2 adding parentheses to CQL WHERE clauses and not appending ALLOW FILTERING.
• Fix ModelType.withinErrorBound() never returning true with error bound zero.
• Fix Apache Spark using the wrong case for predicate push-down of members.
• Fix the denormalized dimensions using the wrong case for the column names.