Passing int should broadcast to fill series in DataFrame

[thomasaarholt]

Describe your feature request

With numpy and pandas you can expect that setting arr[:] = some_int will broadcast the shape of some_int to fill the shape of the dimension being set.

With polars:

import polars as pl
a = {"foo":[1,2,3],}
df = pl.DataFrame(a)
df["foo"] = [1,2,3]
df["foo"] = 1 # <-------- Error is here

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
File ~/mambaforge/envs/geo2/lib/python3.9/site-packages/polars/internals/frame.py:1458, in DataFrame.__setitem__(self, key, value)
   1457 try:
-> 1458     self.replace(key, pli.Series(key, value))
   1459 except Exception:

File ~/mambaforge/envs/geo2/lib/python3.9/site-packages/polars/internals/series.py:227, in Series.__init__(self, name, values, dtype, strict, nan_to_null)
    226 else:
--> 227     raise ValueError("Series constructor not called properly.")

ValueError: Series constructor not called properly.

During handling of the above exception, another exception occurred:

ValueError                                Traceback (most recent call last)
Input In [24], in <module>
      3 df = pl.DataFrame(a)
      4 df["foo"] = [1,2,3]
----> 5 df["foo"] = 1

File ~/mambaforge/envs/geo2/lib/python3.9/site-packages/polars/internals/frame.py:1460, in DataFrame.__setitem__(self, key, value)
   1458         self.replace(key, pli.Series(key, value))
   1459     except Exception:
-> 1460         self.hstack([pli.Series(key, value)], in_place=True)
   1461 # df[["C", "D"]]
   1462 elif isinstance(key, list):

File ~/mambaforge/envs/geo2/lib/python3.9/site-packages/polars/internals/series.py:227, in Series.__init__(self, name, values, dtype, strict, nan_to_null)
    225     self._s = pandas_to_pyseries(name, values)
    226 else:
--> 227     raise ValueError("Series constructor not called properly.")

ValueError: Series constructor not called properly.

[ritchie46]

No it should not. Thanks for the reminder to deprecate DataFrame.__setitem__. ;)

Try this:

df = pl.DataFrame({
    "foo":[1,2,3]
})

df.with_column(pl.lit(1).alias("bar"))

[thomasaarholt]

Hahaha, very well :D