Closed rjzamora closed 3 months ago
Came across a subtle bug/change in dask-expr while debugging a dask-cudf test:
import dask.dataframe as dd df = dd.from_dict({"a": [0, 1]}, 1) df.dtypes.a # dtype('int64') df.a = df.a.astype("float") df.dtypes.a # `meta` is NOT updated: dtype('int64') df["a"] = df.a.astype("float") df.dtypes.a # `meta` IS updated: dtype('float64')
This doesn't work at all fwiw, the column isn't updated either
Came across a subtle bug/change in dask-expr while debugging a dask-cudf test: