[R] Reading a parquet with a Float16 column yields incorrect value

[rodigu]

Describe the bug, including details regarding any error messages, version, and platform.

i created a parquet file in polars where one of the columns has a Float16 type:

import polars as pl

df = pl.DataFrame(
    {"name": "john", "height": 1.78},
    schema={"name": pl.String, "height": pl.Float64},
).write_parquet("./data.parquet")

but reading it with arrow yields a different value:

library(arrow)

df <- read_parquet("./data.parquet")
print(df)

#   name height
# 1 john  16159

Component(s)

R

[raulcd]

In the example you are using pl.Float64, is this supposed to be Float16 instead? Also could you share versions of polars, and Arrow. Can you reproduce with PyArrow or is only R?