rusackas commented on code in PR #41964:
URL: https://github.com/apache/superset/pull/41964#discussion_r3578308852
##########
superset/connectors/sqla/models.py:
##########
@@ -1516,7 +1516,18 @@ def full_name(self) -> str:
def dttm_cols(self) -> list[str]:
l = [c.column_name for c in self.columns if c.is_dttm] # noqa: E741
if self.main_dttm_col and self.main_dttm_col not in l:
- l.append(self.main_dttm_col)
+ # Only treat ``main_dttm_col`` as a datetime column when the
column it
+ # points to is actually temporal. A column whose "Is Temporal"
flag was
+ # removed must not keep being reported as a datetime column just
because
+ # it is still referenced by ``main_dttm_col`` (#30510). When the
column
+ # is not present on the dataset, fall back to the legacy behavior
of
+ # trusting ``main_dttm_col``.
+ main_dttm_column: TableColumn | None = next(
+ (c for c in self.columns if c.column_name ==
self.main_dttm_col),
+ None,
+ )
+ if main_dttm_column is None or main_dttm_column.is_dttm:
+ l.append(self.main_dttm_col)
Review Comment:
Same as I said on the earlier Bito thread: line 1517 above already filters
everything else by the raw `is_dttm` flag, not `is_temporal`, so switching just
this fallback would make `main_dttm_col` more lenient than the rest of
`dttm_cols` for the same NULL-flag case. Keeping the two branches consistent
here.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
