Enables or disables the reader's row-group statistics screen.
With a filter active, the reader reads each row group's footer statistics first and skips the row groups the filter provably cannot match -- no column data is read for those at all. It changes how much of the file is read and nothing else: the rows returned are identical either way, which is exactly what makes an A/B comparison the right test for it and what turning it off is for.
Leaving it on is right for essentially every program. Disable it only to compare the two paths, or if a file's statistics are known to be untrustworthy.
| Type | Intent | Optional | Attributes | Name | ||
|---|---|---|---|---|---|---|
| logical, | intent(in) | :: | enabled |
.true. (the default) lets the reader prune row groups. |
subroutine parquet_set_statistics_prescreen(enabled) logical, intent(in) :: enabled !! .true. (the default) lets the reader prune row groups. cfg_statistics_prescreen = enabled end subroutine parquet_set_statistics_prescreen