|
| 1 | +#### Overview |
| 2 | + |
| 3 | +Apache Arrow is the universal columnar format and multi-language toolbox for fast data interchange and in-memory analytics. |
| 4 | + |
| 5 | +Run-End-Encoding is a composite Arrow datatype that efficiently represents data with many repeated values, such as timeseries. |
| 6 | +While basic support is present, Run-End-Encoding is not universally supported in all Arrow C++ components. |
| 7 | + |
| 8 | +We propose to finish implementing support for Run-End-Encoding in all components of Arrow C++: |
| 9 | + |
| 10 | +* cast kernels (including implicit run-end-encoding or decoding where applicable) |
| 11 | + |
| 12 | +* scalar kernels: |
| 13 | + - all unary kernels (math, etc.) |
| 14 | + - all binary kernels (arithmetic, comparisons, etc.) |
| 15 | + - `coalesce`, `if_else`, `case_when` |
| 16 | + - fix incorrect output in `is_null` and `true_unless_null` |
| 17 | + |
| 18 | +* vector kernels: |
| 19 | + - cumulative aggregations (`cumulative_max`, `cumulative_sum`, etc.) |
| 20 | + - `pairwise_diff` |
| 21 | + - `filter`, `take`, `scatter` |
| 22 | + - `fill_null_forward`, `fill_null_backward` |
| 23 | + - `replace_with_mask` |
| 24 | + - `indices_non_zero` |
| 25 | + - `sort_indices`, `rank`, `rank_normal`, `rank_quantile` |
| 26 | + - `partition_nth_indices` |
| 27 | + - `select_k_unstable` |
| 28 | + - `unique`, `value_counts` |
| 29 | + - `winsorize` |
| 30 | + - fix incorrect output in `drop_null` |
| 31 | + |
| 32 | +* aggregate kernels: |
| 33 | + - `all`, `any` |
| 34 | + - `count_distinct` |
| 35 | + - `approximate_median` |
| 36 | + - `sum`, `product`, `mean`, `mode`, `quantile`, `tdigest` |
| 37 | + - `variance`, `stddev`, `skew`, `kurtosis` |
| 38 | + - `first`, `last`, `min`, `max` |
| 39 | + - `index` |
| 40 | + - fix incorrect output in `count` |
| 41 | + |
| 42 | +* Parquet reader and writer |
| 43 | + |
| 44 | +* CSV reader |
| 45 | + |
| 46 | +* ORC reader and writer |
| 47 | + |
| 48 | +Funders can decide to fund the entire package, or choose the components they are interested in. |
| 49 | + |
| 50 | +##### Are you interested in this project? Either entirely or partially, contact us for more information on how to help us fund it. |
0 commit comments