Arrow 0.5.0 release

[asfimport]

Target date: week of June 12 (or sooner)

Reporter: Wes McKinney / @wesm
Assignee: Wes McKinney / @wesm

Related issues:

• [C++] DictionaryBuilder should use signed integers for indices (is blocked by)
• Some logical types not supported when loading Parquet (is blocked by)
• [Python] Empty "private" directories should be ignored by Parquet interface (is blocked by)
• C++: arrow::test::TestBase::MakePrimitive doesn't fill null_bitmap (is blocked by)
• [Python] Utilize pandas metadata from common _metadata Parquet file if it exists (is blocked by)
• Python: Table.from_pandas doesn't work anymore on partial schemas (is blocked by)
• [Python] Create chunked BinaryArray in Table.from_pandas when a column's data exceeds 2GB (is blocked by)
• [C++] Detect int32 overflow in ListBuilder::Append (is blocked by)
• Getting record batch size with pa.get_record_batch_size returns a size that is too small for pandas DataFrame. (is blocked by)
• [C++] Implement JSON round trip for DictionaryArray (is blocked by)
• Python: Auto-detect nested lists and nested numpy arrays in Pandas (is blocked by)
• [Python] Add convenience function to yield DataFrame from any object that a StreamReader or FileReader can read from (is blocked by)
• Java<->C++ Integration tests for dictionary-encoded vectors (is blocked by)
• [Python] RecordBatchFileReader does not have a schema property (is blocked by)
• [Python] Write support for int96 (is blocked by)

_{Note: This issue was originally created as ARROW-1052. Please see the migration documentation for further details.}