New rule: Decimal with float argument

[kbaskett248]

I'm interested in adding a new rule to check for Decimal being called with a float literal argument. For example:

# Don't
d = Decimal(1.234)
# Do
d = Decimal('1.234')

This seems like a generally useful check. I searched for open issues and PRs involving Decimal and didn't find any mention of this type of check. I'm happy to work on a PR, but wanted to make sure this check makes sense in ruff. Please let me know if I'm missing something or there's any reason this check wouldn't be accepted.

Thanks!

[charliermarsh]

I don't think we have a check for this. Can you say more on why it's something to be avoided?

[mayanyax]

Consider:

>>> Decimal(0.3) == Decimal(0.1) + Decimal(0.1) + Decimal(0.1)
False
>>> Decimal('0.3') == Decimal('0.1') + Decimal('0.1') + Decimal('0.1')
True

In the first example you basically creating an exact representation of the floating point approximation to 0.1. To my mind that is almost certainly not what is wanted as it seems to defeat the purpose of using Decimal in the first place.

[kbaskett248]

Yeah, @mayanyax is exactly right. Passing a float literal could introduce imprecision, which is exactly what we're hoping to avoid.

[mjgp2]

This would be a good addition, I have use cases where this has caused potential bugs.

[Matthieu-LAURENT39]

this is a footgun i keep forgetting, and i would definitely enjoy seeing this rule implemented.

[AlexWaygood]

Fixed by #12909