BUG (2.0rc1): Missing breaking API change in the changelog related to new timedelta64 dtypes

[aberres]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd

x_start = pd.Series(pd.Timestamp(2023,3,14,12))
x_end = pd.Series(pd.Timestamp(2023,3,14,14))

# Pandas 1.5.3
# 0    7200000.0
# dtype: float64
# 
# Pandas 2.0.0rc0
# 0   0 days 02:00:00
# dtype: timedelta64[ms]
(x_end - x_start).astype("timedelta64[ms]")

Issue Description

When test-driving the RC we noticed that generating Plotly graphs failed (snippet for reference below).

When investigating, we found the following in the Plotly code: (x_end - x_start).astype("timedelta64[ms]"). With Pandas < 2 this returned float64 representing the ms between the timestamps.

Now that Pandas 2 has a timedelta64[ms] dtype this dtype is used instead.

When existing code expects a float64 dtype it might break - as in the Plotly case.

As workaround, the following works with all Pandas versions:

(x_end - x_start).astype("timedelta64[ns]") / pd.Timedelta(1, 'ms')`

Maybe it makes sense to add this change to the list of breaking changes?

It might also be worth to mention that casting from np.datetime64[D] results in timedelta64[s] now as noticed here.

Plotly reproduction

import plotly.express as px
import plotly.graph_objects as go
import pandas as pd

df = pd.DataFrame([{"Task": "Job A", "Start": "2022-01-01", "Finish": "2022-01-02"}])
fig_1 = px.timeline(df, x_start="Start", x_end="Finish", y="Task")
fig_1.update_yaxes(autorange="reversed")

Leads to

[...]
File /Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/json/encoder.py:258, in JSONEncoder.iterencode(self, o, _one_shot)
    253 else:
    254     _iterencode = _make_iterencode(
    255         markers, self.default, _encoder, self.indent, floatstr,
    256         self.key_separator, self.item_separator, self.sort_keys,
    257         self.skipkeys, _one_shot)
--> 258 return _iterencode(o, 0)

File ~/venv/weplan3.11/lib/python3.11/site-packages/_plotly_utils/utils.py:136, in PlotlyJSONEncoder.default(self, obj)
    134     except NotEncodable:
    135         pass
--> 136 return _json.JSONEncoder.default(self, obj)

File /Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/json/encoder.py:180, in JSONEncoder.default(self, o)
    161 def default(self, o):
    162     """Implement this method in a subclass such that it returns
    163     a serializable object for ``o``, or calls the base implementation
    164     (to raise a ``TypeError``).
   (...)
    178 
    179     """
--> 180     raise TypeError(f'Object of type {o.__class__.__name__} '
    181                     f'is not JSON serializable')

TypeError: Object of type timedelta is not JSON serializable

Expected Behavior

Changelog mentions the new behavior.

Installed Versions

[aberres]

Ok, ok. I was blind. A colleague pointed me to the correct section.