Add argument to df.to_dicts() and df.to_dict() - maintain_column_order: bool

[DeflateAwning]

Description

It appears that df.to_dict() and to_dicts don't maintain the column order of the keys in the dictionary/dictionaries. This is presumably an optimization.

It would be helpful if you could enforce maintaining the order from the dataframe with an argument.

[cmdlineluser]

Do you happen to have an example?

Anything I've tried so far results in a dict with the same order.

[alexander-beedie]

Likewise; so far I haven't been able to reproduce - and looking at the code it does look like it should always return in-order keys. Definitely need a reproducible example to see if there's an edge-case somehow/somewhere (the keys appear to be inserted in frame-order and Python has had a hard guarantee since 3.6/3.7 that insertion order is maintained) 🤔

[DeflateAwning]

Darn, thanks for checking! Sorry to waste your time.

The actual issue was in the way Pytest compares dicts by taking them out-of-order: https://stackoverflow.com/questions/70023936/pytest-dict-equality-preserving-order-and-nice-diff-output-python-3-7

That's totally my bad; sorry for not vetting the issue better before opening it here.