Hey @Saurabh Mishra Just checking in—were you able to resolve what you described above?

yes! I had patched the

_save_overwrite

behavior to unblock myself.

def _patched_save_overwrite(self, data: DataFrame) -> None:
    """Patched overwrite with schema auto-merge and dynamic partition overwrite.

    Replaces the default ManagedTableDataset._save_overwrite to:
    - Use Delta schema auto-merge instead of overwriteSchema
    - Use dynamic partition overwrite mode (only overwrite partitions in the DataFrame)
    """
    writer = (
        data.write.format(self._table.format)
        .mode("overwrite")
        .option("partitionOverwriteMode", "dynamic")
    )

    if self._table.partition_columns:
        writer.partitionBy(
            *self._table.partition_columns
            if isinstance(self._table.partition_columns, list)
            else [self._table.partition_columns]
        )

    if self._table.location:
        writer.option("path", self._table.location)

    writer.saveAsTable(self._table.full_table_location() or "")

Hi Saurabh, thanks for sharing! Would you like to raise this as a feature request?