BUG: pd.to_datetime() throws if caching is on with Null-like arguments

[anmyachev]

• closes pd.to_datetime() throws if caching is on with Null-like arguments #22305
• tests added / passed
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

The problem was in contradictory behavior between pd.Index([pd.NaT, None]).is_unique that returns False and algorithms.unique([pd.NaT, None]) that returns array([NaT, None], dtype=object) equivalent to the object being unique.

We are only using Index(arg).unique() now.

The test written by @realead was added to verify changes.

[codecov]

Codecov Report

Merging #26078 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #26078      +/-   ##
==========================================
- Coverage   91.89%   91.89%   -0.01%     
==========================================
  Files         175      175              
  Lines       52509    52514       +5     
==========================================
+ Hits        48255    48256       +1     
- Misses       4254     4258       +4

Flag	Coverage Δ
#multiple	90.45% <100%> (ø)	⬆️
#single	40.74% <55.55%> (-0.14%)	⬇️

Impacted Files	Coverage Δ
pandas/core/indexes/base.py	96.78% <100%> (-0.01%)	⬇️
pandas/core/base.py	97.81% <100%> (+0.03%)	⬆️
pandas/io/gbq.py	75% <0%> (-12.5%)	⬇️
pandas/core/frame.py	96.79% <0%> (-0.12%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5a15069...d0ec00c. Read the comment docs.

[codecov]

Codecov Report

Merging #26078 into master will increase coverage by 0.08%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #26078      +/-   ##
==========================================
+ Coverage   91.89%   91.98%   +0.08%     
==========================================
  Files         175      175              
  Lines       52509    52384     -125     
==========================================
- Hits        48255    48184      -71     
+ Misses       4254     4200      -54

Flag	Coverage Δ
#multiple	90.53% <100%> (+0.08%)	⬆️
#single	40.74% <0%> (-0.15%)	⬇️

Impacted Files	Coverage Δ
pandas/core/tools/datetimes.py	84.59% <100%> (ø)	⬆️
pandas/io/gbq.py	75% <0%> (-12.5%)	⬇️
pandas/util/_doctools.py	12% <0%> (-0.88%)	⬇️
pandas/core/panel.py	35.43% <0%> (-0.1%)	⬇️
pandas/core/computation/ops.py	95.62% <0%> (-0.04%)	⬇️
pandas/core/computation/align.py	97.8% <0%> (-0.03%)	⬇️
pandas/core/reshape/melt.py	97.47% <0%> (-0.03%)	⬇️
pandas/core/dtypes/concat.py	96.56% <0%> (-0.02%)	⬇️
pandas/plotting/_core.py	83.76% <0%> (-0.02%)	⬇️
pandas/core/nanops.py	93.82% <0%> (-0.02%)	⬇️
... and 44 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5a15069...a3c66b3. Read the comment docs.

[pep8speaks]

Hello @anmyachev! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2019-04-17 10:56:38 UTC

[jreback]

thanks @anmyachev

[sjvdm]

This issue persists in pandas 1.2.4. Doesn't seem like this was ever merged to prod - are there perhaps any plans?

[gecko17]

I can confirm this issue occurs in pandas 1.2.4. However, it was not reproducible in pandas 1.3.4.
Can anyone confirm this was fixed?