[C++] Stream API: Add support for repeated fields

[asfimport]

The parquet::StreamReader and parquet::StreamWriter classes currently only support required fields.

Support must be added to this API in order for it to be usable when the schema has repeated fields.

 

Reporter: Gawain BOLTON / @gawain-bolton Assignee: Gawain BOLTON / @gawain-bolton

Related issues:

• [C++] Stream API: Add support for optional fields (is cloned by)

_{Note: This issue was originally created as PARQUET-1700. Please see the migration documentation for further details.}

[asfimport]

Wes McKinney / @wesm: Might make sense to think about the API for groups (structs) while also thinking about repeated fields

[asfimport]

Gawain BOLTON / @gawain-bolton: I thought Parquet only supported basic data types?

I will change this ticket to be for repeated fields and create a separate Jira ticket for optional fields as I will have something ready soon to handle these.

[asfimport]

Wes McKinney / @wesm: Parquet indeed supports both structs and arrays/lists

[asfimport]

Micah Kornfield / @emkornfield: it might be nice to have something reusable between Arrow and Parquet for structs (I think at some point we want to add a similar row level iterator API for arrow Tables/RowBatches).