We need documentation describing a migration path from ParquetDataset, at least for the basic user facing API of ParquetDataset (As I read it, that's: construction, projection, filtering, threading, for a first pass). Following this we could mark ParquetDataset as deprecated, building features needed by power users like dask and adding those to the migration document
Reporter: Ben Kietzman / @bkietz
Related issues:
Note: This issue was originally created as ARROW-8047. Please see the migration documentation for further details.
We need documentation describing a migration path from ParquetDataset, at least for the basic user facing API of ParquetDataset (As I read it, that's: construction, projection, filtering, threading, for a first pass). Following this we could mark ParquetDataset as deprecated, building features needed by power users like dask and adding those to the migration document
Reporter: Ben Kietzman / @bkietz
Related issues:
Note: This issue was originally created as ARROW-8047. Please see the migration documentation for further details.