11.37. Release 0.193

General Changes

Fix an infinite loop during planning for queries containing non-trivial predicates.
Fix row_number() optimization that causes query failure or incorrect resultsfor queries that constrain the result of row_number() to be less than one.
Fix failure during query planning when lambda expressions are used in UNNEST or VALUES clauses.
Fix Tried to free more revocable memory than is reserved error for queries that have spilling enabledand run in the reserved memory pool.
Improve the performance of the ST_Contains() function.
Add map_zip_with() lambda function.
Add normal_cdf() function.
Add SET_DIGEST type and related functions.
Add query stat that tracks peak total memory.
Improve performance of queries that filter all data from a table up-front (e.g., due to partition pruning).
Turn on new local scheduling algorithm by default (see Release 0.181).
Remove the informationschema._internal_partitions table.

Apply the authentication methods in the order they are listed in thehttp-server.authentication.type configuration.

Allow dropping any column in a table. Previously, dropping columns otherthan the last one would fail with ConcurrentModificationException.
Correctly write files for text format tables that use non-default delimiters.Previously, they were written with the default delimiter.
Fix reading data from S3 if the data is in a region other than us-east-1.Previously, such queries would fail with"The authorization header is malformed; the region 'us-east-1' is wrong; expecting '<region_name>'",where <region_name> is the S3 region hosting the bucket that is queried.
Enable SHOW PARTITIONS FROM <table> WHERE <condition> to work for tablesthat have more than hive.max-partitions-per-scan partitions as long asthe specified <condition> reduces the number of partitions to below this limit.

Allow renaming tables across schemas. Previously, the target schema was ignored.
Do not allow creating tables in a nonexistent schema.