Currently, for writing a table to disk directly, the defaults are suitable when going to a CSV, IPC, Parquet, etc. However, writing a dataset requires multiple options to be configured, even when defaults could be obvious, e.g., the name for the fragments requires user input, when often they'll be named something like "part{i}.parquet." Ideally, the defaults should be adequate to write a Dataset, without further user configuration.
Reporter: Kae Suarez / @ksuarez1423
Related issues:
Note: This issue was originally created as ARROW-17402. Please see the migration documentation for further details.
Currently, for writing a table to disk directly, the defaults are suitable when going to a CSV, IPC, Parquet, etc. However, writing a dataset requires multiple options to be configured, even when defaults could be obvious, e.g., the name for the fragments requires user input, when often they'll be named something like "part{i}.parquet." Ideally, the defaults should be adequate to write a Dataset, without further user configuration.
Reporter: Kae Suarez / @ksuarez1423
Related issues:
Note: This issue was originally created as ARROW-17402. Please see the migration documentation for further details.