Hi, all. We're proposing Page Filtering in parquet-cpp implementation[1]. Currently, parquet-cpp and arrow only support RowGroup/ColumnChunk level pruning. Now we can support filtering with Parquet PageIndex[2]. The interface can be also used to helping implementing the iceberg positional delete format.
Suggestions/observations from discussion on that draft included: - A RowRanges API in parquet - Support reading RowRanges in PageReader, RecordReader and parquet decoder - Support passing a RowRanges to FileReader in parquet. Sincerely, Xuwei Fu [1] https://docs.google.com/document/d/1SeVcYudu6uD9rb9zRAnlLGgdauutaNZlAaS0gVzjkgM/edit [2] https://github.com/apache/parquet-format/blob/master/PageIndex.md