What a click on the table lands on: the row on screen, counted from the top (none
on the header), and the column (none on the rail or the row numbers).
What the fill-reading worker needs to make it the buffer: the adjoining rows on
hand, the view and the caps, as planned. A trim that would copy (see
trim_rows) runs here, off the UI thread; apply_async_collect installs the
result as is.
What an open learned besides frame and schema, given once via
DataTableState::with_open so count, row groups, files and notes agree. Defaults
mean not found.
A dataset of many files and how to read only some: remote, or a local Hive
directory once every footer is known. Polars reads files in order (row 900,000
means reading every file before it); with each file’s rows known, a buffer scans
just the files holding its rows.
How a remote dataset reads some of its files, as the pass behind an open found
them. Together because they describe one list: the scan is built at the new
schema, and the counter answers one entry per file, which must match urls or
its answer is dropped.
The view as it stood before a query or view replaced it: a checkpoint. A query can
plan and still fail on its rows (a value that will not cast); then the table returns
to this, rows and all. Frames and buffers are shared, not copied. Taken by
DataTableState::rollback_point or DataTableState::try_transition, restored by
DataTableState::roll_back.
Builds a scan of some of a dataset’s files as the full scan reads them, the second
argument’s columns read as text from every file instead of the majority type.
Reads every footer of a dataset opened from a couple of them; None when they
could not be read (the dataset stays as opened). Blocks, reporting to the progress.