Reshaping into invalid shapes has inconsistent error behavior #24807
Description
Checks
- I have checked that this issue has not already been reported.
- I have confirmed this bug exists on the latest version of Polars.
Reproducible example
df = pl.DataFrame({"a": list(range(2 * 3 * 2))})
# This fails, because Polars cannot fill up to a window of 12
q = df.lazy().select(pl.col("a").reshape((12, 3)))
print(f"{q.collect(engine="in-memory") = }")
print(f"{q.collect(engine="streaming") = }")
# This does not fail, even though we have too many rows (4, not 2)
q = df.lazy().select(pl.col("a").reshape((2, 3)))
print(f"{q.collect(engine="in-memory") = }")
print(f"{q.collect(engine="streaming") = }")Log output
Issue description
The error behavior of Expr.reshape() is inconsistent.
- In case the resulting number of rows is too small to fill a single "reshape window", then Polars will raise an error.
- In case the resulting number of rows is bigger than the "reshape window", Polars will ignore this and give back a dataframe with too many rows.
Expected behavior
Expr.reshape() should give an error if the number of rows does not exactly match the number of rows specified in the first reshape dimension.
Installed versions
--------Version info---------
Polars: 1.34.0
Index type: UInt32
Platform: Linux-6.14.0-33-generic-x86_64-with-glibc2.39
Python: 3.12.3 (main, Aug 14 2025, 17:47:21) [GCC 13.3.0]
Runtime: unknown
----Optional dependencies----
Azure CLI <not installed>
adbc_driver_manager 1.8.0
altair 5.5.0
azure.identity 1.25.0
boto3 1.40.18
cloudpickle 3.1.1
connectorx 0.4.4
deltalake 1.1.4
fastexcel 0.15.1
fsspec 2025.9.0
gevent 25.8.2
google.auth 2.40.3
great_tables 0.18.0
matplotlib 3.10.6
numpy 2.2.6
openpyxl 3.1.5
pandas 2.3.2
polars_cloud 0.1.4
pyarrow 21.0.0
pydantic 2.11.9
pyiceberg 0.10.0
sqlalchemy 2.0.43
torch <not installed>
xlsx2csv 0.8.4
xlsxwriter 3.2.8