Alexey Kudinkin
f1e3762a94
[HUDI-2950] Addressing performance traps in Bulk Insert/Layout Optimization (#4234)
* Cleaned up Z-curve/Hilbert ordering seqs:
- Streamlined flow
- Removed unnecessary operations (double-mapping, boxing, etc)
Updated `CollectionUtils::combine` to avoid AL resizing
* Tidying up
* Reducing small objects churn due to Scala/Java conversions by re-using `RowFactory`, passing `Object[]`
* Fixing name resolution (disambiguation overloads)
* `lint`
* Replaced `OverwriteAvroPayloadRecord` w/ `RewriteRecordPayload` to avoid unnecessary Avro ser/de loop
* Added `PathCachingFileName` to avoid fetching substrings every time file-name is fetched;
Inject `PathCachingFileName` into `HoodieWrapperFileSystem.convertPathWithScheme`
* Drastically reducing size of the `ArrayDeque` allocated by `ObjectSizeCalculator`
* XXX
* Missing license
* Fixed refs (after rebase)
* Fixing compilation failure in Scala 2.11
* `PathCachingFileName` > `FileNameCachingPath`
* Tidying up