chore: Fix Decimal precision annotation

[vyasr]

On construction the precision if set to None is shifted to 38, so the precision is always non-None.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.89%. Comparing base (a3474bb) to head (ba95966).
⚠️ Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #25227      +/-   ##
==========================================
+ Coverage   81.77%   81.89%   +0.11%     
==========================================
  Files        1713     1714       +1     
  Lines      236780   239028    +2248     
  Branches     3013     3249     +236     
==========================================
+ Hits       193629   195745    +2116     
- Misses      42381    42485     +104     
- Partials      770      798      +28     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[alexander-beedie]

Thanks for contributing - however, as per the docstring, precision=None means "inferred". So, while any eventual DataFrame will indeed be initialised with a specific (inferred) Decimal precision (that will typically be 38), it is fine for the dtype itself to have precision=None. The typing as-is looks correct to me.

pl.Decimal(precision=None, scale=3)
# Decimal(precision=None, scale=3)

[TomAugspurger]

@alexander-beedie are we looking at different versions? I see this:

In [1]: import polars as pl

In [2]: pl.Decimal(precision=None, scale=3)
Out[2]: Decimal(precision=38, scale=3)

In [3]: pl.__version__
Out[3]: '1.35.2'

It looks like #24742 might have changed the docstring, but not the type annotation.

[alexander-beedie]

Oof; I stand corrected - apparently I was looking at this on a secondary laptop that hadn't been updated to latest main 🤦‍♂️ Reopening and will merge shortly - thanks for catching me on that!

[coastalwhite]

I don't think this is correct. None now means maximum, which is 38. As decimals were stabilized last release, this would be a breaking change.

[coastalwhite]

Nevermind, this only touches the member variables and not the constructor.

[vyasr]

@alexander-beedie thanks for reviewing so quickly! However, I think the behavior you're showing might be outdated. #24542 modified how precision works for decimals (specifically this diff), see the extended discussion in #19784. As a result precision is always set to a non-None value (of 38) on construction if None is provided. Consider:

❯ uv pip install polars==1.33
...
❯ python -c "import polars as pl; print(pl.Decimal(precision=None, scale=3))"
Decimal(precision=None, scale=3)
❯ uv pip install polars==1.34
...
❯ python -c "import polars as pl; print(pl.Decimal(precision=None, scale=3))"

[vyasr]

Ah my browser just refreshed and I see multiple new comments and that you all already figured this out as well and merged the PR, so we're all set. Thanks all!