Matthijs Brobbel created PARQUET-2299:
-----------------------------------------
Summary: Use `true` instead of `1` as default value for
`is_compressed bool` field
Key: PARQUET-2299
URL: https://issues.apache.org/jira/browse/PARQUET-2299
Project: Parquet
Issue Type: Improvement
Components: parquet-format
Reporter: Matthijs Brobbel
I noticed that the default value for the optional boolean `is_compressed` field
of the `DataPageHeaderV2` struct has a default value of `1`. According to the
Thrift docs a boolean value is either `true` or `false`.
This currently works because the Apache Thrift compiler internally handles
bools as ints:
-
https://github.com/apache/thrift/blob/3880a09565a9a1dad028b3679746eafac268c819/compiler/cpp/src/thrift/thriftl.ll#L208-L209
-
https://github.com/apache/thrift/blob/3880a09565a9a1dad028b3679746eafac268c819/compiler/cpp/src/thrift/main.cc#L748
It may however not work with other Thrift compilers that are more strict about
this.
Based on the docs and a test
(https://github.com/apache/thrift/blob/3880a09565a9a1dad028b3679746eafac268c819/test/ThriftTest.thrift#L406)
in the Thrift repository it seems that using `true` here is the correct way of
defining a default for an optional bool field.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
