Question on parquet-protobuf and parquet-pig

Hi,

I have a question about protobuf to parquet conversion.
I have a message like this (simplified) :
```Java

message MyMessage {
repeated string language = 1;
}
```

parquet-protobuf convert it to the following schema :

```Java
message MyMessage {
repeated binary language (UTF8);
}
```

But, according to TestPigSchameConverter.java, the correct schema should be :

```Java
message MyMessage {
optional group language (LIST) {
repeated binary value (UTF8);
}
}
```

Language is an optional list of language, i want to store zero or more language.

Who have the correct schema for my case ? parquet-protobuf or parquet-pig ?

**Reporter**: [Christian Nguyen Van Than](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=cnguyen@nextperf.com)

_{**Note**: *This issue was originally created as [PARQUET-354](https://issues.apache.org/jira/browse/PARQUET-354). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

Compare the protobuf conversion behavior with the expected schema in TestPigSchameConverter.java, starting with how parquet-protobuf and parquet-pig represent repeated string fields. Determine whether the schemas differ incorrectly for zero-or-more values, and document a reproducible conclusion or test that resolves which representation is correct.

Written by the indexing model from the issue text.