Jakob Blomer
Jakob Blomer
Addresses several bugs on big-endian architecture, some in the testing code only, some in the actual code. Tested manually on Power8/EL7. Also tested that RNTuple data written on little-endian can...
Implement two improvements to the RNTuple RDF scheduler for the multi-threaded case: 1. Files of a chain are opened in batches the background: while a batch of files (batch size...
The file page sink, on commit, collects the streamer info records of custom classes. These records are passed together with the records from unsplit fields to the file writer, which...
### Explain what you would like to see improved and how. The RNTupleParallelWriter should be extended to allow to fix the logical order of clusters written to disk. This would...
### Explain what you would like to see improved and how. The RClusterPool should be improved - to work with a memory budget rather than a fixed number of clusters...
### Explain what you would like to see improved and how. Currently, `RNTupleDescriptor::CreateModel()` always creates a full model with a default entry. There can be situations in which a bare...
### Explain what you would like to see improved and how. The boolean template in `RNTupleView` steers whether or not the view is owned by the user. It would be...
Moves from fixed page sizes on write to adaptive page sizes, following the original idea of @hahnjo The new mechanism is explained in the tuning.md document in the PR. The...
### Explain what you would like to see improved and how. Implement the RNTuple standard benchmarks using ORC: https://orc.apache.org/ Measure read/write time, memory consumption, and final file size. Moved from...