[PIG-3067] HBaseStorage should be split up to become more manageable

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Patch Info: Patch Available
• Assignee: Christoph Bauer
• Reporter: Christoph Bauer
• Created: 23/Nov/12 16:27
• Updated: 12/Aug/13 16:32

Description

HBaseStorage has become quite big (>1100 lines).

I propose to split it up into more managable parts. I believe it will become a lot easier to maintain.

I split it up like this:

HBaseStorage

• settings:LoadStoreFuncSettings
  • options
  • caster
  • udfProperties
  • contextSignature
  • columns:ColumnInfo - moved to its own class-file
• loadFuncDelegate:HBaseLoadFunc - LoadFunc implementation
  • settings:LoadStoreFuncSettings (s.a.)
  • scanner:HBaseLoadFuncScanner - everything scan-specific
  • tupleIterator:HBaseTupleIterator - interface for public Tuple getNext()
• storeFuncDelegate:HBaseStorFunc - StorFunc implementation
  • settings:LoadStoreFuncSettings (s.a.)

Issue Links

• relates to: PIG-2921 Provide a bulkloadable option in HBaseStorage