Details
-
Sub-task
-
Status: Resolved
-
Low
-
Resolution: Fixed
-
None
-
None
Description
After CASSANDRA-9932, there is only one instance where this class is needed, and that is SSTableIterator. It would be much simpler if, like in AbstractBTreePartition, the slices were passed into the SSTableIterator on construction. This would make the control flow easier to follow both:
- logically, because:
- memtables and sstables would have the same access pattern; and
- our SSTableIterator would not have two modes of (parallel) operation, which is kind of confusing (the slices are independent of the non-sliced iteration)
- in the debugger, because we would have one fewer wrapping iterator