Upgrade DataFusion to version 56 by nuno-faria · Pull Request #1826 · apache/datafusion-python

nuno-faria · 2026-11-07T16:22:40Z

Which issue does this PR close?

N/A.

Rationale for this change

A new DataFusion version is being released: apache/datafusion#24461

What changes are included in this PR?

There are new types like Show (which replaced DescribeTable), AsOfJoin, and MaterializedCte, as well as some other minor changes. I will add them as comments.

Are there any user-facing changes?

New DF version. Other than that the main breaking change is that DescribeTable has been removed. I added a note to the upgrade guide.

nuno-faria · 2026-11-07T16:23:29Z

apache/datafusion#23829

nuno-faria · 2026-11-07T16:23:44Z

apache/datafusion#25112

nuno-faria · 2026-11-07T16:23:58Z

apache/datafusion#25112

nuno-faria · 2026-11-07T16:24:39Z

    }

-    /// Retrieves the right input `LogicalPlan` to this `Join` node
+    /// Retrieves the equality keys of this `Join` node


Noticed when creating the ASOF join that the prev comment was wrong.

nuno-faria · 2026-11-07T16:24:50Z

apache/datafusion#26269

nuno-faria · 2026-11-07T16:27:00Z

        &self,
        _ctx: &dyn Session,
-        projection: Option<&Vec<usize>>,
+        projection: Option<&[usize]>,


apache/datafusion#24376

nuno-faria · 2026-11-07T16:27:50Z

@@ -5385,7 +5385,7 @@ def map_entries(map: Expr) -> Expr:
 def element_at(map: Expr, key: Expr) -> Expr:
    """Returns the value for a given key in the map.

-    Returns ``[None]`` if the key is absent.
+    Returns ``[]`` if the key is absent.


apache/datafusion#24999

nuno-faria · 2026-11-07T16:28:34Z


 def test_sql_with_options_no_statements(ctx):
-    sql = "SET time zone = 1;"
+    sql = "SET time zone = 'UTC';"


Timezone is now validated.

nuno-faria · 2026-11-07T16:30:03Z

@@ -1811,7 +1811,7 @@ def test_read_csv_with_options(tmp_path, as_read, global_ctx):
        [
            pa.array(["Alice", "Bob", "Phil", "Karin"]),
            pa.array([30.0, 25.0, 75.0, 50.0]),
-            pa.array(["New York; NY", None, "Detroit' MI", "Stockholm\nSweden"]),
+            pa.array(["New York; NY", "", "Detroit' MI", "Stockholm\nSweden"]),


This now appears as an empty string since it does not match the null_regex. From apache/datafusion#25254

nuno-faria · 2026-11-07T16:30:20Z

+    with pytest.raises(
+        Exception, match=r"Invalid compression level \d+ for \w+\. Expected a level"
+    ):


apache/datafusion#25626

nuno-faria · 2026-11-07T17:45:08Z

@timsaucer I think the problem with the _test_three_library_query_planner comes from the FFI crate upstream:

Exception: DataFusion error: FFI error: SanityCheckPlan: the plan produced by physical optimizer
rule 'EnsureRequirements' does not satisfy the order and distribution requirements

caused by

Error during planning:

Plan:
  AggregateExec: mode=FinalPartitioned, gby=[numbers.A % Int64(2)@0 as numbers.A % Int64(2)], aggr=[count(Int64(1))]
    FFI_ExecutionPlan: RepartitionExec, number_of_children=1
      AggregateExec: mode=Partial, gby=[CAST(A@0 AS Int64) % 2 as numbers.A % Int64(2)], aggr=[count(Int64(1))]
        FFI_ExecutionPlan: DataSourceExec, number_of_children=0

does not satisfy distribution requirements: KeyPartitioned[[numbers.A % Int64(2)@0]])

Child-0 output partitioning: RoundRobinBatch(4)

It appears that the problem is now shown by apache/datafusion#26461.

andygrove · 2026-11-14T10:39:29Z

@nuno-faria you were right that the _test_three_library_query_planner failure comes from upstream. I filed apache/datafusion#27427 and opened a fix in apache/datafusion#27428.

FFI_ExecutionPlan never carried a plan's input requirements: input distribution, required ordering, maintains_input_order. When the host's optimizer rules run across FFI on the planner library's plan, EnsureRequirements can't see that FinalPartitioned needs hash-partitioned input, so it adds only a round-robin repartition. On 55, SanityCheckPlan also ran as a host-side rule and was just as blind, so the test passed. apache/datafusion#26461 moved the check into DefaultPhysicalPlanner, so 56 rejects the plan.

I cherry-picked the fix onto the branch-56 commit this PR pins (0f3b5a08) and built this branch against it. The query-planner example suite passes (51/51, including test_spawning_plan_across_three_libraries), and so does the FFI example suite (59/59). Once the fix lands on branch-56 and the pin here is bumped, this test should pass without changes on our side.

andygrove

This is an AI generated review. Feel free to ignore any points that you do not agree with.

Blocking

test_spawning_plan_across_three_libraries fails because of an upstream FFI bug. FFI_ExecutionPlan doesn't carry input requirements across the boundary. Tracked in apache/datafusion#27427, with a fix in apache/datafusion#27428. With the fix applied to branch-56, this PR passes all FFI example tests (51/51 and 59/59).
CI's TPC-H step will fail next. tpchgen-cli 4.0 removed --format. Here the step was skipped only because the FFI step failed first. #1832 fixes it, and this PR will need to rebase onto it.
Dependency source. [patch.crates-io] points at a moving git branch-56, and the comment above it says to remove it before a release candidate. For now it could bump to the current branch-56 tip (cce6374e6, which includes the 56.0.0 version bump). I ran the full suite against that tip with the FFI fix applied, and it is clean: main suite with doctests, both FFI example suites, TPC-H 22/22 and clippy. Once 56.0.0 is on crates.io it should switch to that.

Should fix

Show.schema() returns None for DESCRIBE <table> (crates/core/src/expr/show.rs). Previously DescribeTable.schema() returned the table's schema in this case. Part of this comes from upstream: ShowKind::Columns now holds only table_ref and detail, with no schema. But the binding exposes no table_ref and none of the other kinds' payloads (variable name, function filter, etc.), so users can't get the table back. The upgrade guide shows kind() == "columns" but doesn't mention that schema() is now None. It could suggest ctx.table(name).schema() instead.
Some user-visible behavior changes are missing from upgrade-guides.md. The tests and docstrings were updated for these, but the guide wasn't:
- map_extract / element_at with a missing key now returns [] instead of [None].
- CSV null_regex now also applies to the data, not only to schema inference (apache/datafusion#25254). Fields that don't match the regex, including the empty fields of truncated rows, come back as "" instead of null. This looks like intended upstream behavior and the test update is correct, but it is worth a note.
- Time zone values in SET are now validated.
The new bindings aren't covered from Python. test_asof_join and test_materialized_cte only compare display_indent() output, which is upstream's Rust Display. No test reaches AsOfJoin, MaterializedCte, MaterializedCteScan, CreateExternalCatalog or DropCatalog through to_variant() or their accessors (match_condition(), on(), etc.). A wrong match arm or a broken accessor in PyLogicalPlan::to_variant would still pass.
crates/core/src/expr/describe_table.rs is orphaned. Its mod line was removed, but the file is still tracked and still references the removed upstream DescribeTable. It can be deleted.
MSRV. The workspace has rust-version = "1.88", but DataFusion 56 declares rust-version = "1.95.0" (and rust-toolchain.toml is now 1.99.0). It should be raised to at least 1.95.

Minor

The Python pin also applies to Windows. In .github/workflows/build.yml, uv sync --python 3.14 is in the job whose matrix is os: [macos-latest, windows-latest], so the Windows abi3 build is pinned too, while the commit message says macOS only. The pin is also repeated in two jobs.
Show(schema, output_schema) can only build ShowKind::Describe, and kind() returns bare strings. That's fine for now, but a docstring note listing the possible kind() values would help users who match on them.

Upgrade DataFusion to version 56

5eb35f9

nuno-faria commented Nov 7, 2026

View reviewed changes

nuno-faria added 2 commits November 7, 2026 16:33

Fix lint

0061384

Pin macOS abi3 Python version to 3.14

b1b413a

nuno-faria mentioned this pull request Nov 9, 2026

Release DataFusion 56.0.0 (Oct/Nov 2026) apache/datafusion#24461

Open

40 of 61 tasks

Xuanwo mentioned this pull request Nov 11, 2026

Tracking issue for geodatafusion 0.7.0 / Python 0.5.0 release (DataFusion 56 & Arrow 60) datafusion-contrib/geodatafusion#163

Open

1 of 19 tasks

This was referenced Nov 14, 2026

FFI_ExecutionPlan loses input distribution and ordering requirements, so optimizer rules across FFI produce invalid plans apache/datafusion#27427

Open

fix: carry input requirements across FFI_ExecutionPlan apache/datafusion#27428

Open

andygrove mentioned this pull request Nov 14, 2026

ci: use tpchgen-cli 4.x parquet subcommand #1832

Open

andygrove reviewed Nov 14, 2026

View reviewed changes

Conversation

nuno-faria commented Nov 7, 2026

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are there any user-facing changes?

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!

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!

nuno-faria commented Nov 7, 2026

Uh oh!

andygrove commented Nov 14, 2026

Uh oh!

andygrove left a comment

Choose a reason for hiding this comment

Blocking

Should fix

Minor

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants