Constructor reorders columns and rows for (lists or dicts) of dicts

[soerenwolfers]

Code Sample

import pandas as pd
print(pd.DataFrame({0:{'B':'First row','A':'Second row'}}))
print(pd.DataFrame([{'B':'First column','A':'Second column'}]))

Results in

            0
A  Second row
B   First row
               A             B
0  Second column  First column

Problem description

The two constructor calls above reorder the rows or columns, respectively, alphabetically, so the row or column, respectively, indexed by 'A' always comes first.

Expected Output

The order of the dict should be preserved now that Python dictionaries are officially ordered.
Hence, the row or column, respectively, indexed by 'A' should come second.

I don't need to make an argument for this, because pandas itself has already decided so:
https://pandas.pydata.org/pandas-docs/version/0.23.0/whatsnew.html#whatsnew-0230-api-breaking-dict-insertion-order

Versions

Tested with pandas 0.24.2 and Python 3.7

[WillAyd]

This is a duplicate of quite a few issues. The problem with this is that there is no such thing as guaranteed ordering in two dimensions.

[WillAyd]

Closing as a duplicate of #24859

[soerenwolfers]

@WillAyd I don't understand

There is no such thing as guaranteed ordering in two dimensions.

Do you mean

There may not be a unique ordering if multiple dicts are passed in.

If so, what prevents pandas from using the unique order if it exists?