feat: Add `Series.from_iterable` by dangotbanned · Pull Request #2933 · narwhals-dev/narwhals

dangotbanned · 2025-08-03T13:31:15Z

Closes #2925

What type of PR is this? (check all applicable)

Related issues

Supersedes enh: deprecate nw.new_series in favour of Series.from_iterable #2642
Part of API: io functions for v2 #2116

Checklist

Code follows style guide (ruff)
Tests added
Documented the changes

Tasks

Closes #2925

resolves https://results.pre-commit.ci/run/github/760058710/1754227876.clY_LXANTgmVEVkE2dlDMw

tests/series_only/from_iterable_test.py

https://github.com/narwhals-dev/narwhals/actions/runs/16705863358/job/47283854415?pr=2933

dangotbanned · 2025-08-03T14:26:43Z

I feel like we should be issuing a warning here at-a-minimum

narwhals/narwhals/_pandas_like/utils.py

Lines 504 to 511 in 1cfcafb

    
           if isinstance_or_issubclass(dtype, dtypes.Datetime): 
        
               # Pandas does not support "ms" or "us" time units before version 2.0 
        
               if implementation is Implementation.PANDAS and backend_version < ( 
        
                   2, 
        
               ):  # pragma: no cover 
        
                   dt_time_unit = "ns" 
        
               else: 
        
                   dt_time_unit = dtype.time_unit

But really I'd want an exception raised and an update to the nw.Datetime docs 🤔

Update

Moved to #2964

#2933 (comment)

dangotbanned · 2025-08-03T14:56:22Z

narwhals/stable/v2/__init__.py

    def _dataframe(self) -> type[DataFrame[Any]]:
        return DataFrame

+    # TODO @dangotbanned: Fix `from_numpy` override missing in `v2` in another PR


Note

Resolve missing v2 method(s) and tests in another PR

narwhals/series.py

Resolves #2933 (comment)

Going to parametrize more for `test_series_from_iterable` #2933 (comment)

#2933 (comment)

- pola-rs/polars#22757 - https://github.com/narwhals-dev/narwhals/blob/992515f6238296c3af5323b29ab0a537c1fcdbaa/narwhals/_arrow/utils.py#L84-L91

Fixes https://github.com/narwhals-dev/narwhals/actions/runs/16726722613/job/47344204297?pr=2933

#2933 (comment)

Hopefully fix https://github.com/narwhals-dev/narwhals/actions/runs/16728654716/job/47350971315?pr=2933

https://github.com/narwhals-dev/narwhals/actions/runs/16728654717/job/47350971274?pr=2933

`0.20.7` is close enough to our minimum and this conversion I can see being more common than the rest

https://github.com/narwhals-dev/narwhals/actions/runs/16729965116/job/47355391903?pr=2933

dangotbanned · 2025-08-08T15:53:14Z

Thanks for the review @MarcoGorelli 😍

this from_iterable test takes nearly 3 seconds on my laptop, which I feel is quite a lot, is it possible to trim it down slightly?

Sure I could take a look!

Would you be able to run pytest --durations=0 on that module please?
I'm just hoping to have the biggest impact on your setup

They aren't showing in my top 100 slowest:

Show durations

>>> pytest --durations=100
====== slowest 100 durations ===
2.10s call     tests/tpch_q1_test.py::test_q1[dask]
1.35s call     tests/frame/schema_test.py::test_nested_dtypes_dask
1.27s call     tests/v1_test.py::test_gather_every_dask_v1[2-1]
1.12s call     tests/no_imports_test.py::test_dask
1.11s call     tests/v1_test.py::test_gather_every_dask_v1[1-2]
0.98s call     tests/dependencies/imports_test.py::test_to_native_namespace_min_version[dask]
0.85s call     tests/namespace_test.py::test_preserve_type_var[dask]
0.67s call     tests/expr_and_series/over_test.py::test_over_unsupported_dask
0.61s call     tests/new_series_test.py::test_new_series_dask
0.48s call     tests/translate/from_native_test.py::test_eager_only_lazy_dask[False-context0]
0.44s call     tests/frame/join_test.py::test_anti_join[sqlframe-join_key0-filter_expr0-expected0]
0.43s call     tests/expr_and_series/rolling_mean_test.py::test_rolling_mean_expr_lazy_ungrouped[sqlframe-expected_a1-2-2-False]
0.42s call     tests/frame/join_test.py::test_join_duplicate_column_names[sqlframe]
0.42s call     tests/expr_and_series/rank_test.py::test_lazy_rank_expr_desc[sqlframe-data1-max]
0.41s call     tests/frame/lazy_test.py::test_lazy_backend[pandas-dask1]
0.41s call     tests/expr_and_series/quantile_test.py::test_quantile_expr[ibis-lower-expected0]
0.40s call     tests/expr_and_series/diff_test.py::test_diff_lazy[sqlframe]
0.39s call     tests/expr_and_series/rolling_sum_test.py::test_rolling_sum_expr_lazy_grouped[sqlframe-expected_a5-4-1-True]
0.37s call     tests/expr_and_series/is_first_distinct_test.py::test_is_first_distinct_expr_lazy_grouped[sqlframe]
0.36s call     tests/expr_and_series/operators_test.py::test_comparand_operators_expr[sqlframe-__gt__-expected5]
0.36s call     tests/modern_polars/unpivot_test.py::test_unpivot[sqlframe]
0.35s call     tests/expr_and_series/name/to_uppercase_test.py::test_to_uppercase_anonymous[sqlframe]
0.34s call     tests/v1_test.py::test_toplevel
0.34s call     tests/expr_and_series/rolling_mean_test.py::test_rolling_mean_expr_lazy_grouped[sqlframe-expected_a1-2-2-False]
0.33s call     tests/expr_and_series/binary_test.py::test_expr_binary[sqlframe]
0.33s call     tests/frame/join_test.py::test_anti_join[pandas[pyarrow]-join_key2-filter_expr2-expected2]
0.33s call     tests/expr_and_series/name/prefix_test.py::test_suffix_after_alias[sqlframe]
0.33s call     tests/expr_and_series/exp_test.py::test_exp_expr[sqlframe]
0.33s call     tests/system_info_test.py::test_get_deps_info
0.32s call     tests/frame/select_test.py::test_select_duplicates[ibis]
0.32s call     tests/frame/join_test.py::test_inner_join_single_key[sqlframe]
0.32s call     tests/expr_and_series/shift_test.py::test_shift_lazy[sqlframe]
0.32s call     tests/dtypes_test.py::test_2d_array[sqlframe]
0.31s call     tests/frame/join_test.py::test_joinasof_time[pandas[pyarrow]-forward-expected1]
0.31s call     tests/v2_test.py::test_with_version[sqlframe]
0.31s call     tests/expr_and_series/rank_test.py::test_lazy_rank_expr[sqlframe-data0-dense]
0.31s call     tests/frame/join_test.py::test_inner_join_two_keys[sqlframe]
0.31s call     tests/expr_and_series/null_count_test.py::test_null_count_expr[sqlframe]
0.30s call     tests/expr_and_series/cum_sum_test.py::test_lazy_cum_sum_grouped[sqlframe-True-expected_a1]
0.30s call     tests/selectors_test.py::test_datetime[polars[eager]]
0.30s call     tests/expr_and_series/cum_sum_test.py::test_lazy_cum_sum_ordered_by_nulls[sqlframe-False-expected_a0]
0.30s call     tests/translate/from_native_test.py::test_eager_only_pass_through_main[ibis-True-True-context2]
0.28s call     tests/frame/to_native_test.py::test_to_native[sqlframe]
0.28s call     tests/expr_and_series/rolling_sum_test.py::test_rolling_sum_expr_lazy_ungrouped[sqlframe-expected_a0-2-None-False]
0.27s call     tests/expr_and_series/mean_test.py::test_expr_mean_expr[sqlframe-expr1]
0.27s call     tests/frame/explode_test.py::test_explode_single_col[sqlframe-l3-expected_values1]
0.26s call     tests/frame/join_test.py::test_left_join_overlapping_column[sqlframe]
0.26s call     tests/expr_and_series/all_horizontal_test.py::test_allh[sqlframe]
0.25s call     tests/expr_and_series/operators_test.py::test_comparand_operators_scalar_expr[ibis-__gt__-expected5]
0.25s call     tests/expr_and_series/exclude_test.py::test_exclude[ibis-exclude_selector2-expected_cols2]
0.25s call     tests/expr_and_series/str/zfill_test.py::test_str_zfill[sqlframe]
0.24s call     tests/expr_and_series/clip_test.py::test_clip_expr[ibis-0-4-expected1]
0.22s call     tests/frame/explode_test.py::test_explode_single_col[sqlframe-l2-expected_values0]
0.22s call     tests/expr_and_series/name/to_uppercase_test.py::test_to_uppercase[ibis]
0.22s call     tests/system_info_test.py::test_get_sys_info
0.22s call     tests/expr_and_series/fill_null_test.py::test_fill_null_limits[sqlframe]
0.21s call     tests/expr_and_series/clip_test.py::test_clip_expr[sqlframe-None-4-expected2]
0.21s call     tests/expr_and_series/all_horizontal_test.py::test_allh_iterator[ibis]
0.21s call     tests/expr_and_series/rolling_mean_test.py::test_rolling_mean_expr_lazy_ungrouped[ibis-expected_a5-4-1-True]
0.20s call     tests/expr_and_series/null_count_test.py::test_null_count_expr[ibis]
0.19s call     tests/expr_and_series/operators_test.py::test_comparand_operators_expr[sqlframe-__ne__-expected1]
0.19s call     tests/frame/select_test.py::test_alias_invalid[sqlframe]
0.19s call     tests/system_info_test.py::test_show_versions
0.18s call     tests/expr_and_series/rank_test.py::test_lazy_rank_expr_desc[duckdb-data0-average]
0.18s call     tests/expr_and_series/exclude_test.py::test_exclude[sqlframe-exclude_selector0-expected_cols0]
0.18s call     tests/v2_test.py::test_with_version[ibis]
0.17s call     tests/frame/join_test.py::test_left_join[sqlframe]
0.17s call     tests/expr_and_series/rank_test.py::test_rank_expr_in_over_desc[ibis-max]
0.17s call     tests/expr_and_series/cum_sum_test.py::test_lazy_cum_sum_grouped[ibis-True-expected_a1]
0.16s call     tests/expr_and_series/operators_test.py::test_logic_operators_expr_scalar[duckdb-__ror__-expected3]
0.16s call     tests/translate/from_native_test.py::test_eager_only_pass_through_main[sqlframe-False-False-context0]
0.16s call     tests/modern_polars/unpivot_test.py::test_unpivot[ibis]
0.16s call     tests/expr_and_series/binary_test.py::test_expr_binary[ibis]
0.16s call     tests/expr_and_series/exp_test.py::test_exp_expr[ibis]
0.16s call     tests/frame/group_by_test.py::test_fancy_functions[sqlframe]
0.16s call     tests/expr_and_series/is_first_distinct_test.py::test_is_first_distinct_expr_lazy[ibis]
0.16s call     tests/expr_and_series/operators_test.py::test_comparand_operators_scalar_expr[ibis-__le__-expected2]
0.15s call     tests/expr_and_series/str/zfill_test.py::test_str_zfill[ibis]
0.15s call     tests/expr_and_series/rolling_sum_test.py::test_rolling_sum_expr[polars[eager]]
0.15s call     tests/expr_and_series/shift_test.py::test_shift_lazy_grouped[ibis]
0.15s call     tests/expr_and_series/diff_test.py::test_diff_lazy_grouped[ibis]
0.15s call     tests/expr_and_series/name/prefix_test.py::test_suffix_after_alias[ibis]
0.14s call     tests/selectors_test.py::test_datetime[pandas]
0.14s call     tests/dtypes_test.py::test_cast_decimal_to_native
0.13s call     tests/expr_and_series/binary_test.py::test_expr_binary[duckdb]
0.13s call     tests/expr_and_series/dt/truncate_test.py::test_truncate_tz_aware_duckdb
0.13s call     tests/expr_and_series/mean_test.py::test_expr_mean_expr[ibis-expr0]
0.13s call     tests/expr_and_series/rank_test.py::test_lazy_rank_expr_desc[sqlframe-data1-ordinal]
0.13s call     tests/expr_and_series/dt/truncate_test.py::test_truncate[sqlframe-1d-expected6]
0.13s call     tests/expr_and_series/exp_test.py::test_exp_expr[duckdb]
0.13s call     tests/frame/to_native_test.py::test_to_native[ibis]
0.12s call     tests/read_scan_test.py::test_scan_csv[sqlframe]
0.12s call     tests/expr_and_series/cast_test.py::test_cast[sqlframe]
0.12s call     tests/expr_and_series/cast_test.py::test_cast_datetime_tz_aware[pandas[pyarrow]]
0.12s call     tests/v1_test.py::test_dask_order_dependent_ops
0.12s call     tests/dtypes_test.py::test_2d_array[ibis]
0.12s call     tests/expr_and_series/fill_null_test.py::test_fill_null_strategies_with_limit_as_none[sqlframe]
0.12s call     tests/expr_and_series/operators_test.py::test_logic_operators_expr[duckdb-__or__-expected1]
0.11s call     tests/read_scan_test.py::test_scan_parquet_kwargs
0.11s call     tests/expr_and_series/shift_test.py::test_shift_lazy[duckdb]
======= 8402 passed, 359 skipped, 509 xfailed in 22.81s

dangotbanned · 2025-08-08T18:03:10Z

#2933 (comment)

@MarcoGorelli

is it possible to trim it down slightly?

So I went ahead and did some experimenting locally:

Show diff

diff --git a/tests/series_only/from_iterable_test.py b/tests/series_only/from_iterable_test.py
index 733c0728c..17f7fb105 100644
--- a/tests/series_only/from_iterable_test.py
+++ b/tests/series_only/from_iterable_test.py
@@ -21,6 +21,7 @@ if TYPE_CHECKING:
         ValuesView,
     )
 
+    from _pytest.mark import ParameterSet
     from typing_extensions import TypeAlias
 
     from narwhals._namespace import EagerAllowed
@@ -29,6 +30,10 @@ if TYPE_CHECKING:
     IntoIterable: TypeAlias = Callable[..., Iterable[Any]]
 
 
+def param_slow(*args: object) -> ParameterSet:
+    return pytest.param(*args, marks=pytest.mark.slow)
+
+
 class UserDefinedIterable:
     def __init__(self, iterable: Iterable[Any]) -> None:
         self.iterable: Iterable[Any] = iterable
@@ -53,7 +58,7 @@ def dict_values(iterable: Iterable[Any]) -> ValuesView[Any]:
     return dict(enumerate(iterable)).values()
 
 
-_INTO_ITER_3RD_PARTY: list[IntoIterable] = []
+_INTO_ITER_3RD_PARTY: list[IntoIterable | ParameterSet] = []
 
 if find_spec("numpy"):  # pragma: no cover
     import numpy as np
@@ -64,7 +69,7 @@ else:  # pragma: no cover
 if find_spec("pandas"):  # pragma: no cover
     import pandas as pd
 
-    _INTO_ITER_3RD_PARTY.extend([pd.Index, pd.array, pd.Series])
+    _INTO_ITER_3RD_PARTY.extend([pd.Index, param_slow(pd.array), pd.Series])
 else:  # pragma: no cover
     ...
 if find_spec("polars"):  # pragma: no cover
@@ -83,16 +88,20 @@ if find_spec("pyarrow"):  # pragma: no cover
 else:  # pragma: no cover
     ...
 
-_INTO_ITER_STDLIB: tuple[IntoIterable, ...] = (
+_INTO_ITER_STDLIB: tuple[IntoIterable | ParameterSet, ...] = (
     list,
     tuple,
     iter,
-    deque,
-    generator_function,
+    param_slow(deque),
+    param_slow(generator_function),
     generator_expression,
 )
-_INTO_ITER_STDLIB_EXOTIC: tuple[IntoIterable, ...] = dict_keys, dict_values
-INTO_ITER: tuple[IntoIterable, ...] = (
+
+_INTO_ITER_STDLIB_EXOTIC: tuple[ParameterSet, ...] = (
+    param_slow(dict_keys),
+    param_slow(dict_values),
+)
+INTO_ITER: tuple[IntoIterable | ParameterSet, ...] = (
     *_INTO_ITER_STDLIB,
     *_INTO_ITER_STDLIB_EXOTIC,
     UserDefinedIterable,
@@ -113,7 +122,7 @@ def _ids_into_iter(obj: Any) -> str:
 @pytest.mark.parametrize(
     ("values", "dtype"),
     [
-        ((4, 1, 2), nw.Int32),
+        param_slow((4, 1, 2), nw.Int32),
         ((-1, 5, 100), None),
         ((2.1, 2.7, 2.0), nw.Float64),
         (("one", "two"), nw.String),
@@ -167,10 +176,10 @@ def test_series_from_iterable(
 
 @pytest.mark.parametrize(("values", "expected_dtype"), [((4, 1, 2), nw.Int64)])
 def test_series_from_iterable_infer(
-    eager_backend: EagerAllowed, values: Sequence[Any], expected_dtype: IntoDType
+    eager_implementation: EagerAllowed, values: Sequence[Any], expected_dtype: IntoDType
 ) -> None:
     name = "b"
-    result = nw.Series.from_iterable(name, values, backend=eager_backend)
+    result = nw.Series.from_iterable(name, values, backend=eager_implementation)
     assert result.dtype == expected_dtype
     assert_equal_series(result, values, name)
 
@@ -182,14 +191,14 @@ def test_series_from_iterable_not_eager() -> None:
         nw.Series.from_iterable("", [1, 2, 3], backend=backend)
 
 
-def test_series_from_iterable_numpy_not_1d(eager_backend: EagerAllowed) -> None:
+def test_series_from_iterable_numpy_not_1d(eager_implementation: EagerAllowed) -> None:
     pytest.importorskip("numpy")
     import numpy as np
 
     with pytest.raises(ValueError, match="only.+1D numpy arrays"):
-        nw.Series.from_iterable("", np.array([[0], [2]]), backend=eager_backend)
+        nw.Series.from_iterable("", np.array([[0], [2]]), backend=eager_implementation)
 
 
-def test_series_from_iterable_not_iterable(eager_backend: EagerAllowed) -> None:
+def test_series_from_iterable_not_iterable(eager_implementation: EagerAllowed) -> None:
     with pytest.raises(TypeError, match="iterable.+got.+int"):
-        nw.Series.from_iterable("", 2000, backend=eager_backend)  # type: ignore[arg-type]
+        nw.Series.from_iterable("", 2000, backend=eager_implementation)  # type: ignore[arg-type]

I can reproduce ~3s using pytest "tests/series_only/from_iterable_test.py" --numprocesses=1
No individual case is consistently slower than others
- There's just a lot of them! (200ish)😅
Arbitrarily marking half as slow only brings down the timing to 2.59s

So besides just outright removing cases or randomly generating them - the best option would be running with --numprocesses=logical (which I'm guessing you are already?)

Something else we could try is adapting this into a hypothesis test?
I'm thinking we could mark the entire thing as slow, but then use some strategies to randomly sample constructors + DTypes
Maybe as a sub-issue on (#2959) or (#2818) as a follow-up?

It could be good to use this as a chance to get wider coverage - in addition to just making the testing feedback loop shorter 🙂

dangotbanned · 2025-08-12T12:55:07Z

narwhals/_polars/series.py

+            iterable = cast("Sequence[Any]", data)
+        native = pl.Series(name=name, values=iterable)
+        if dtype:
+            native = native.cast(narwhals_to_native_dtype(dtype, version))
        return cls.from_native(native, context=context)


I'm thinking this change is what's causing #2933 (comment)

Well great, (revert: try not downcasting polars) did indeed fix tubular ci

But there's some version branching needed for at least polars==0.20.21

https://github.com/narwhals-dev/narwhals/actions/runs/16909565353/job/47907363225?pr=2933

@MarcoGorelli I think this was fixed in this polars PR you reviewed

fix(python): Respect dtype and strict in pl.Series's constructor for pyarrow arrays, numpy arrays, and pyarrow-backed pandas pola-rs/polars#15962

If so, should I re-add the cast fix I had for polars<0.20.26 only - or is an xfail preferred?

In (fix: backport respect dtype, optimize branching) - I'm taking the view that Series.from_iterable/nw.new_series is a critical entrypoint.

So the aim is to both:

Make sure it has wide version-compatibility

Add as little overhead as possible, especially on the happy path

The fix in (48cf17b) may have been the source of (#2933 (comment))

Trying to keep this entry point as lightweight as possible, while being reliable across versions #2933 (comment)

FBruzzesi

Thanks @dangotbanned - I left only one comment regarding the tests run time.

One question I have (which might be written somewhere but I missed it): are we planning to replace the code in new_series with this eventually? Everything (except the trial to see if UNKNOWN implementation which I commented below) is the same as _new_series_impl

FBruzzesi · 2025-08-13T20:26:19Z

narwhals/_polars/series.py

+        else:  # pragma: no cover
+            if (not SERIES_ACCEPTS_PD_INDEX) and is_pandas_index(values):
+                values = values.to_series()
+            native = pl.Series(name, values)
+            if dtype_pl:
+                native = native.cast(dtype_pl)


Oh boy 😩

I tried something new here when thinking about (#2933 (comment)), which might not be visible from the diff alone

Keeps the compat code fast (we evaulate this once)

Doesn't clutter the impl with comments

Can use markdown to explain the problem + link to issue/PR

FBruzzesi · 2025-08-13T20:43:36Z

tests/series_only/from_iterable_test.py

Trying to find a compromise here to Marco's comment on these tests running for too long. To run half of them, one idea is to use only eager_implementation. At the end of the day it does not really matter how we instantiate the Implementation, and using eager_backend just duplicates the number of tests.

If anything, we should test somewhere else that Implementation.from_backend(Implementation.PANDAS) and Implementation.from_backend("pandas") return the same implementation. Once that guaranteed (which I am pretty sure it is), we can run half of these tests

Thanks @FBruzzesi

Did you know that I've been doing that in this PR since (efc5c4e)?

I added eager_implementation in this PR see diff, and I agree that we should use it more elsewhere 🙂

My bad! There is still one use of eager_backend but it's for a small number of tests anyway

No worries!

Btw, I am open to reducing the test runtime and investigated a little in (#2933 (comment))

My preference is to defer this to another issue - but if you see it as blocking - then I can work on it here?

narwhals/series.py

dangotbanned · 2025-08-14T10:04:50Z

are we planning to replace the code in new_series with this eventually?

Absolutely!

(#2933 (comment)) should cover most of it

I guess the plan in my head is:

Add all of (API: io functions for v2 #2116), which only contain the working parts
Address (Establish safe patterns using Implementation.UNKNOWN #2786), which will be even easier after (feat(typing): Add IntoBackend alias #2971) 😉
Migrate the fixed nw.new_series, nw.from_numpy, etc code into these new constructors

So we'll then have this and the extension point will work everywhere 🥳

from typing import Any

from narwhals._utils import Version, Implementation
from narwhals.typing import IntoBackend
from narwhals.series import Series

some_version = Version.MAIN


def new_series(
    name: str,
    values: Any,
    dtype: IntoDType | None = None,
    *,
    backend: IntoBackend,  # <-- or whatever we end up annotating
) -> Series[Any]:
    return some_version.series.from_iterable(name, values, dtype, backend=backend)

FBruzzesi · 2025-08-16T19:57:09Z

@dangotbanned @MarcoGorelli what's left in this PR? Is the concern only on test speed? Marco maybe run it once again while not live streaming, I can expect the results to be much different 😂

MarcoGorelli

thanks both, happy to ship it!

dangotbanned · 2025-08-17T16:26:29Z

Thank you both for getting this over the line

#2933

Fixes #2933 (comment)

dangotbanned added 5 commits August 3, 2025 12:26

feat: Add Series.from_iterable

2afc8c8

Closes #2925

feat: Add to v1, v2

ba7a983

docs: Adapt Series.from_numpy

4841d9a

test: Port new_series tests

5c28080

test: cover failure cases

fdc108b

dangotbanned added the enhancement New feature or request label Aug 3, 2025

docs: Add to api reference

d939ef7

resolves https://results.pre-commit.ci/run/github/760058710/1754227876.clY_LXANTgmVEVkE2dlDMw

dangotbanned changed the title ~~feat: Add `Series.from_iterable~~ feat: Add Series.from_iterable Aug 3, 2025

dangotbanned commented Aug 3, 2025

View reviewed changes

tests/series_only/from_iterable_test.py Outdated Show resolved Hide resolved

dangotbanned added 2 commits August 3, 2025 13:56

test: cover v1, v2

02ab3cb

test: don't rely on pandas supporting [us]

1cfcafb

https://github.com/narwhals-dev/narwhals/actions/runs/16705863358/job/47283854415?pr=2933

test: xfail pandas [ms]

2a9ac28

#2933 (comment)

dangotbanned commented Aug 3, 2025

View reviewed changes

dangotbanned marked this pull request as ready for review August 3, 2025 14:58

Merge branch 'main' into series-from-iterable

b6e2895

EdAbati reviewed Aug 4, 2025

View reviewed changes

narwhals/series.py Outdated Show resolved Hide resolved

dangotbanned added 2 commits August 4, 2025 12:16

Merge remote-tracking branch 'upstream/main' into series-from-iterable

2137178

refactor: fix RET505

28da5a0

Resolves #2933 (comment)

This comment was marked as resolved.

Sign in to view

dangotbanned added 3 commits August 4, 2025 12:55

test: Split out infer case

7fd179c

Going to parametrize more for `test_series_from_iterable` #2933 (comment)

test: Check lots of iterables

efc5c4e

#2933 (comment)

test(DRAFT): xfail polars w/ pd.array, todo pyarrow fix

328ecf0

- pola-rs/polars#22757 - https://github.com/narwhals-dev/narwhals/blob/992515f6238296c3af5323b29ab0a537c1fcdbaa/narwhals/_arrow/utils.py#L84-L91

dangotbanned marked this pull request as draft August 4, 2025 14:59

dangotbanned added 6 commits August 4, 2025 16:22

fix: Ensure cast occurs before pa.chunked_array call

3f13379

Fixes https://github.com/narwhals-dev/narwhals/actions/runs/16726722613/job/47344204297?pr=2933

test: Cover other data types

bd2a084

#2933 (comment)

fix: Try downcasting in polars

48cf17b

Hopefully fix https://github.com/narwhals-dev/narwhals/actions/runs/16728654716/job/47350971315?pr=2933

cov plz

9bfd3c4

https://github.com/narwhals-dev/narwhals/actions/runs/16728654717/job/47350971274?pr=2933

fix: backport pd.Index support

c666872

`0.20.7` is close enough to our minimum and this conversion I can see being more common than the rest

cov

4a27299

https://github.com/narwhals-dev/narwhals/actions/runs/16729965116/job/47355391903?pr=2933

dangotbanned added 2 commits August 8, 2025 20:11

Merge branch 'main' into series-from-iterable

48cb0b3

Merge branch 'main' into series-from-iterable

ebfeb1b

This comment was marked as resolved.

Sign in to view

Merge branch 'main' into series-from-iterable

773fcbb

dangotbanned commented Aug 12, 2025

View reviewed changes

dangotbanned added 5 commits August 12, 2025 12:57

revert: try *not* downcasting polars

cd98af6

The fix in (48cf17b) may have been the source of (#2933 (comment))

fix: backport respect dtype, optimize branching

4400166

Trying to keep this entry point as lightweight as possible, while being reliable across versions #2933 (comment)

Merge remote-tracking branch 'upstream/main' into series-from-iterable

f73a2f9

Merge branch 'main' into series-from-iterable

f0d1a4f

Merge branch 'main' into series-from-iterable

3fc9e06

FBruzzesi reviewed Aug 13, 2025

View reviewed changes

dangotbanned added 2 commits August 15, 2025 08:49

Merge branch 'main' into series-from-iterable

8b31c92

Merge branch 'main' into series-from-iterable

a30f54f

MarcoGorelli approved these changes Aug 17, 2025

View reviewed changes

dangotbanned merged commit f57991c into main Aug 17, 2025
31 checks passed

dangotbanned deleted the series-from-iterable branch August 17, 2025 16:26

dangotbanned added a commit that referenced this pull request Aug 17, 2025

fix: integrate Series.from_iterable

666eff6

#2933

dangotbanned added a commit that referenced this pull request Aug 18, 2025

fix: Add missing v2 constructor overrides

b8acd29

Fixes #2933 (comment)

This was referenced Aug 18, 2025

fix: Add missing v2 constructor overrides #3006

Merged

Establish safe patterns using Implementation.UNKNOWN #2786

Open

MarcoGorelli removed the fix label Aug 20, 2025

dangotbanned mentioned this pull request Sep 6, 2025

feat: Warn on incompatible TimeUnit in pandas<2 #3007

Merged

9 tasks

dangotbanned mentioned this pull request Sep 21, 2025

feat: Add {nw,DataFrame}.from_dicts #3148

Merged

11 tasks

This was referenced Oct 11, 2025

fix(DONT MERGE): duckdb>=1.4.1 typing & warnings #3189

Closed

fix: Always accept {Expr,Series}.is_in(other: Iterable) #3207

Open

Conversation

dangotbanned commented Aug 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What type of PR is this? (check all applicable)

Related issues

Checklist

Tasks

Uh oh!

Uh oh!

dangotbanned commented Aug 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Update

Uh oh!

dangotbanned Aug 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

This comment was marked as resolved.

dangotbanned commented Aug 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dangotbanned commented Aug 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as resolved.

Choose a reason for hiding this comment

Uh oh!

dangotbanned Aug 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

FBruzzesi left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

dangotbanned commented Aug 14, 2025

Uh oh!

FBruzzesi commented Aug 16, 2025

Uh oh!

MarcoGorelli left a comment

Choose a reason for hiding this comment

Uh oh!

dangotbanned commented Aug 17, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

dangotbanned commented Aug 3, 2025 •

edited

Loading

dangotbanned commented Aug 3, 2025 •

edited

Loading

dangotbanned Aug 3, 2025 •

edited

Loading

dangotbanned commented Aug 8, 2025 •

edited

Loading

dangotbanned commented Aug 8, 2025 •

edited

Loading

dangotbanned Aug 12, 2025 •

edited

Loading