You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Issue #1: you can't filter on that column because (my theory, based on the evidence) the ScanNode takes a projection and filter, but the filter is not evaluated with the augmented schema, so it doesn't find __filename. This seems fixable in C++.
Proof that it is in the ScanNode: If we collapse() the query after projecting to include filename but before the filter, the filter doesn't get included in the ScanNode, it's only applied after, as a FilterNode. This works:
Issue #2, following on that: you can only add the augmented fields at the start of the query, something that goes in the ScanNode. This seems like something we would have to catch in R and error at the time add_filename() is called. That could probably be covered in ARROW-17356.
We can project to add augmented fields like
__filename
, but there are a few catches. Given:Issue #1: you can't filter on that column because (my theory, based on the evidence) the ScanNode takes a projection and filter, but the filter is not evaluated with the augmented schema, so it doesn't find
__filename
. This seems fixable in C++.Proof that it is in the ScanNode: If we
collapse()
the query after projecting to include filename but before the filter, the filter doesn't get included in the ScanNode, it's only applied after, as a FilterNode. This works:A related failure mode: you have to first project to include the augmented column, you can't just include it in a filter:
Issue #2, following on that: you can only add the augmented fields at the start of the query, something that goes in the ScanNode. This seems like something we would have to catch in R and error at the time add_filename() is called. That could probably be covered in ARROW-17356.
Reporter: Neal Richardson / @nealrichardson
Note: This issue was originally created as ARROW-18286. Please see the migration documentation for further details.
The text was updated successfully, but these errors were encountered: