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
{{ message }}
Repository navigation
Commit 7497946
Browse filesBrowse the repository at this point in the historyBrowse files
fix: look up projected Iceberg field ids by exact name in CometScanRule
With spark.sql.caseSensitive=false the rule matched projected columns to
field ids through the session resolver, so in a table with columns a and
A both resolved to a's id. The nested-field check then skipped A, and on
Spark 3.5 and 4.0 a field renamed under A read back as NULL from files
written before the rename. Look ids up by exact name, as the serde does.
0 commit comments