aminghadersohi commented on code in PR #45358:
URL: https://github.com/apache/superset/pull/45358#discussion_r4300282405
##########
tests/unit_tests/mcp_service/chart/test_bullet_tool_paths.py:
##########
@@ -1611,12 +1615,12 @@ async def
test_bullet_timestamp_categories_from_dataframe_reach_fastmcp(
"-0.0010000000000287557",
"1704153600000",
"null",
- "1729989000123.456",
+ str(dublin_epoch_ms),
Review Comment:
Verified at 2f61887b363d92563d77084d5ad87fc231c2977a. Reverting this oracle
to `str()` fails
`test_bullet_dateutil_categories_preserve_chart_data_selected_instants[0]` with
`'1729989000000' != '1729989000000.0'` while `[123456]` passes, and mutating
`_dateutil_named_offset_without_hooks` still kills all eight cases — the offset
axis stays independent. No other `str()`-on-float oracle remains in these two
files.
One measured caveat on the formatting axis: the expectation now shares
production's helper, so regressing `_javascript_number_string` to `str()`
leaves `test_bullet_chart.py`'s two cases and both `[*-ascii]` cells green —
the ASCII branch asserts `category[:20] in content`, and `1729989000000` is a
prefix of `1729989000000.0`. Only `[*-vega_lite]` and
`test_bullet_number_categories_match_node_ieee754_string_semantics` fail, so
the regression is still caught, just not by the new `microsecond=0` cells.
##########
tests/unit_tests/mcp_service/dataset/tool/test_query_dataset.py:
##########
@@ -497,6 +497,19 @@ async def
test_query_dataset_normalizes_supported_producer_timezones(
dataset = _make_dataset(
columns=[_make_column("created_at", is_dttm=True)],
)
+ dublin_fold: datetime = datetime(
+ 2024,
+ 10,
+ 27,
+ 1,
+ 30,
+ 0,
+ 123456,
+ tzinfo=dateutil_tz.gettz("Europe/Dublin"),
+ fold=1,
Review Comment:
Verified at 2f61887b363d92563d77084d5ad87fc231c2977a. Measured on this head:
dateutil returns +01:00 for both fold values at this instant, while ZoneInfo
returns +01:00 and +00:00; swapping the two expected strings fails both added
cases, so the ZoneInfo fixture genuinely selects on fold rather than restating
a constant offset. The coverage lands as parametrized cases of the existing
test rather than a new test function. The remaining dateutil `fold=1` fixtures
under `tests/unit_tests/mcp_service` pin offset reconstruction rather than fold
selection, so they stay inert by design.
--
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]
