Describe the bug
Direct pl.Array pass-through works with in-memory executor after #23773, but nullable Array fails when in-memory DataFrame is split into multiple streaming partitions.
With currently supported Polars 1.42.1, slicing Array containing outer nulls produces Arrow C layout that double-counts slice offset, parent retains nonzero offset while child is already sliced. libcudf correctly rejects this layout.
Polars fixed underlying export issue in 28602, first included in Polars 1.43.2., but cudf-polars currently supports polars>=1.35,<1.43.
Steps/Code to reproduce bug
import polars as pl
print("Polars:", pl.__version__)
query = pl.LazyFrame(
{
"embedding": pl.Series(
[
[0.0, 1.0],
[2.0, None],
None, # Outer null in the second partition.
[3.0, 4.0],
],
dtype=pl.Array(pl.Float32, 2),
)
}
)
engine = pl.GPUEngine(
executor="streaming",
raise_on_fail=True,
executor_options={
"max_rows_per_partition": 2,
"fallback_mode": "raise",
"dynamic_planning": None,
},
)
print("CPU result:")
print(query.collect())
print("Streaming GPU result:")
print(query.collect(engine=engine))Outer null is in the nonzero-offset second partition. Single partition succeeds.
Output
The second partition fails while being converted from Polars through Arrow:
ValueError: CUDF failure at: .../from_arrow_host.cu:
fixed-size-list child is shorter than its parent layout requires
Expected behavior
Query completes on streaming GPU executor, preserves values and nulls, and returns Array(Float32, shape=(2,)).
Environment details
Click here to see environment details
**git***
commit 0e44ceba6365ba04a4fb47ef2de2ad9d771737bc (grafted, HEAD -> fix/cudf-polars-nullable-array-streaming, origin/fix/cudf-polars-nullable-array-streaming)
Author: 0guban0v <[email protected]>
Date: Thu Sep 10 23:35:33 2026 -0700
Fix nullable Array streaming scans
**git submodules***
***OS Information***
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=22.04
DISTRIB_CODENAME=jammy
DISTRIB_DESCRIPTION="Ubuntu 22.04.5 LTS"
PRETTY_NAME="Ubuntu 22.04.5 LTS"
NAME="Ubuntu"
VERSION_ID="22.04"
VERSION="22.04.5 LTS (Jammy Jellyfish)"
VERSION_CODENAME=jammy
ID=ubuntu
ID_LIKE=debian
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
UBUNTU_CODENAME=jammy
Linux instance-5ndnjfzx-main 6.17.0-35-generic #35~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Tue May 26 19:30:42 UTC 2 x86_64 x86_64 x86_64 GNU/Linux
***GPU Information***
Fri Sep 11 07:00:47 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 610.43.02 KMD Version: 610.43.02 CUDA UMD Version: 13.3 |
+-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA RTX A6000 Off | 00000000:07:00.0 Off | Off |
| 30% 25C P8 25W / 300W | 1MiB / 49140MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+
***CPU***
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 64
On-line CPU(s) list: 0-5
Off-line CPU(s) list: 6-63
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) CPU E5-2683 v4 @ 2.10GHz
CPU family: 6
Model: 79
Thread(s) per core: 2
Core(s) per socket: 16
Socket(s): 2
Stepping: 1
CPU max MHz: 3000.0000
CPU min MHz: 1200.0000
BogoMIPS: 4199.64
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 pti intel_ppin ssbd ibrs ibpb stibp tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm rdt_a rdseed adx smap intel_pt xsaveopt cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts vnmi md_clear flush_l1d
Virtualization: VT-x
L1d cache: 1 MiB (32 instances)
L1i cache: 1 MiB (32 instances)
L2 cache: 8 MiB (32 instances)
L3 cache: 80 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-15,32-47
NUMA node1 CPU(s): 16-31,48-63
Vulnerability Gather data sampling: Not affected
Vulnerability Ghostwrite: Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit: KVM: Mitigation: Split huge pages
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Old microcode: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP conditional; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsa: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Vmscape: Mitigation; IBPB before exit to userspace
***CMake***
***g++***
/usr/bin/g++
g++ (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
***nvcc***
***Python***
/home/ubuntu/cudf_pr_cu12/bin/python
Python 3.12.13
***Environment Variables***
PATH : /home/ubuntu/cudf_pr_cu12/bin:/home/ubuntu/.local/bin:/home/ubuntu/.vscodium-server/bin/4c0b0c6cc561d2d3636d1ec250935431876ce4dc/bin/remote-cli:/home/ubuntu/.local/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin
LD_LIBRARY_PATH :
NUMBAPRO_NVVM :
NUMBAPRO_LIBDEVICE :
CONDA_PREFIX :
PYTHON_PATH :
conda not found
***pip packages***
/home/ubuntu/cudf_pr_cu12/bin/pip
Package Version
------------------------ ---------------------------
cuda-bindings 12.9.7
cuda-core 1.2.0
cuda-pathfinder 1.8.1
cuda-toolkit 12.9.2.0
cudf-polars-cu12 26.12.0a54.post260911053105
cudf-streaming-cu12 26.12.0a54.post260911053105
cupy-cuda12x 14.2.0
kvikio-cu12 26.12.0a4.post260911050846
libcudf-cu12 26.12.0a54.post260911053105
libcudf-streaming-cu12 26.12.0a54.post260911053105
libkvikio-cu12 26.12.0a4.post260911050846
librapidsmpf-cu12 26.12.0a8.post260910164721
librmm-cu12 26.12.0a3.post260911050837
libucx-cu12 1.21.0
libucxx-cu12 0.53.0a6.post260911053052
numpy 2.5.3
nvidia-cublas-cu12 12.9.2.10
nvidia-cuda-nvrtc-cu12 12.9.86
nvidia-cuda-runtime-cu12 12.9.79
nvidia-cufft-cu12 11.4.1.4
nvidia-cufile-cu12 1.14.1.1
nvidia-curand-cu12 10.3.10.19
nvidia-cusolver-cu12 11.7.5.82
nvidia-cusparse-cu12 12.5.10.65
nvidia-libnvcomp-cu12 5.3.0.16
nvidia-ml-py 13.610.43
nvidia-nvjitlink-cu12 12.9.86
nvtx 0.2.16
packaging 26.3
pip 26.2.1
polars 1.42.1
polars-runtime-32 1.42.1
pyarrow 25.0.1
pylibcudf-cu12 26.12.0a54.post260911053105
rapids-logger 0.3.0
rapidsmpf-cu12 26.12.0a8.post260910164721
rmm-cu12 26.12.0a3.post260911050837
typing_extensions 4.16.0
ucxx-cu12 0.53.0a6.post260911053052
Additional context
Source checkout shown by print_env.sh was used only to run env-report script. The failure was reproduced with installed cudf-polars-cu12 wheel listed above.
Describe the bug
Direct
pl.Arraypass-through works with in-memory executor after #23773, but nullable Array fails when in-memory DataFrame is split into multiple streaming partitions.With currently supported Polars 1.42.1, slicing Array containing outer nulls produces Arrow C layout that double-counts slice offset, parent retains nonzero offset while child is already sliced. libcudf correctly rejects this layout.
Polars fixed underlying export issue in 28602, first included in Polars 1.43.2., but cudf-polars currently supports polars>=1.35,<1.43.
Steps/Code to reproduce bug
Outer null is in the nonzero-offset second partition. Single partition succeeds.
Output
The second partition fails while being converted from Polars through Arrow:
Expected behavior
Query completes on streaming GPU executor, preserves values and nulls, and returns Array(Float32, shape=(2,)).
Environment details
Click here to see environment details
Additional context
Source checkout shown by
print_env.shwas used only to run env-report script. The failure was reproduced with installedcudf-polars-cu12wheel listed above.