Spark CSV does not support storing BinaryType columns by default.
Data source implementation that handles the writing of files and tables to a filesystem.
Data source implementation that handles the writing of files and tables to a filesystem.
By default Spark is used to write tables and the Hadoop filesystem configured in Spark is used to write files. The file/folder/table structure into which the graphs are stored is defined in DefaultGraphDirectoryStructure.
Spark CSV does not support storing BinaryType columns by default. This data source implementation encodes BinaryType columns to Hex-encoded strings and decodes such columns back to BinaryType. This feature is required because ids within CAPS are stored as BinaryType.