Details
-
Sub-task
-
Status: Accepted
-
Minor
-
Resolution: Unresolved
-
None
-
None
-
None
-
Minor
Description
This should improve efficiency and memory usage, and may allow pointer comparison to check if two actual outputs are identical.
It is non trivial because
i) the memory buffer passed is sometimes part of a stream, with no known size.
ii) the grouping information is separate, which is a pain for the hash table
richardkchapman something to discuss...
The primary benefit would be to common up definitions from the clients (especially roxie), rather than dafilesrv.
Another possible change is to base the grouped meta data implementation on the ungrouped, but that wouldn't immediately help the comparison.