Unsupported Features

The following features from Presto 0.101 may work but are not officially supported:

• Installing presto via the tar ball

• Running presto via the launcher script

• Queue configuration

• The following connectors:

  • Cassandra
  • Kafka
  • PostgreSQL
  • MySQL
  • Hive-hadoop1
  • Hive-cdh4

• ROW data type

• Developer guide, including:

  • SPI
  • Pluggable functions

Known Limitations

• The SQL keyword end is used as a column name in system.runtime.queries, so in order to query from that column, end must be wrapped in quotes.
• DECIMAL and NUMERIC datatypes are not supported
• FULL OUTER JOIN is not supported
• NATURAL JOIN is not supported
• SEMI JOIN is not supported.
• Scalar subqueries are not supported – e.g. WHERE x = (SELECT y FROM ...)
• Correlated subqueries are not supported
• Non-equi joins are only supported for inner join – e.g. "n_name" < "p_name"
• EXISTS, EXCEPT, and INTERSECT are not supported.
• DROP TABLE IF EXISTS is not supported.
• ROLLUP is not supported

Hive Datatype Notes

There are a few caveats for using the Hive connector, in particular related to datatype support.

Columns with the following types are not visible in a table:

• DECIMAL
• DECIMAL with parentheses
• CHAR

Some types are only partially supported.

• VARCHAR is not supported in ORC and RCBinary file formats.

Additionally, Presto does not map Hive data types 1-to-1.

• All integral types are mapped to BIGINT
• FLOAT and DOUBLE are mapped to DOUBLE
• STRING and VARCHAR are mapped to VARCHAR

Usually, this should not be a problem, but the mapping could be visible if column values are passed to Hive UDFs, or through slight differences in mathematical functions.

The granularity of the TIMESTAMP datatype is to milliseconds in Presto, while Hive supports microseconds

TIMESTAMP values that do not have an explicit timezone are parsed according to Hive’s hive.time-zone property in the Hive catalog definition. When results are printed to the console, the Presto client’s JVM timezone is used.

TIMESTAMP, DATE and BINARY datatypes are not supported for the Parquet file type.