[Variant] Add `VariantBuilder` values check by sdf-jkl · Pull Request #10016 · apache/arrow-rs

sdf-jkl · 2026-05-25T01:33:13Z

Which issue does this PR close?

Closes [Variant] VariantBuilder (wrongly?) accepts 0+ variant values #7870.

Rationale for this change

(Single) VariantBuilder accepts 0/1+ values on finish.

What changes are included in this PR?

Add only 1 value validation for VariantBuilder
Move try_new_list/try_new_object to VariantBuilder
Replaced VariantBuilder with WritableMetadataBuilder in tests that only build metadata.

Are these changes tested?

Yes, existing unit tests pass.
Ran benchmarks locally and added some [#inline]

Are there any user-facing changes?

…tBuidlerExt` trait

…tbuilder-count-check

sdf-jkl · 2026-05-28T22:09:37Z

@alamb @scovich please take a look

alamb

Makes sense to me -- thank you @sdf-jkl

fyi @scovich and @klion26

sdf-jkl · 2026-06-09T16:42:47Z

@alamb could you run some benchmarks?

#9791 (comment)

run benchmarks variant_builder variant_kernels variant_validation

alamb · 2026-06-09T17:50:48Z

run benchmarks variant_builder variant_kernels variant_validation

adriangbot · 2026-06-09T17:53:57Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4662479980-508-f6l2k 6.12.68+ #1 SMP Sat May 2 07:49:07 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           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; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing variantbuilder-count-check (f372e3a) to 1377761 (merge-base) diff
BENCH_NAME=variant_builder
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench variant_builder
BENCH_FILTER=
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-09T17:53:58Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4662479980-510-mvmlz 6.12.68+ #1 SMP Sat May 2 07:49:07 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           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; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing variantbuilder-count-check (f372e3a) to 1377761 (merge-base) diff
BENCH_NAME=variant_validation
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench variant_validation
BENCH_FILTER=
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-09T17:54:39Z

🤖 Arrow criterion benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4662479980-509-4msnv 6.12.68+ #1 SMP Sat May 2 07:49:07 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           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; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing variantbuilder-count-check (f372e3a) to 1377761 (merge-base) diff
BENCH_NAME=variant_kernels
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench variant_kernels
BENCH_FILTER=
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-06-09T17:56:09Z

🤖 Arrow criterion benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           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; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

group                               main                                   variantbuilder-count-check
-----                               ----                                   --------------------------
bench_validate_complex_object       1.00    182.7±5.36µs        ? ?/sec    1.00    182.1±5.20µs        ? ?/sec
bench_validate_large_nested_list    1.00     19.2±0.46ms        ? ?/sec    1.00     19.3±0.45ms        ? ?/sec
bench_validate_large_object         1.00     52.2±1.12ms        ? ?/sec    1.00     52.2±1.09ms        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	30.0s
Peak memory	12.5 MiB
Avg memory	8.1 MiB
CPU user	27.4s
CPU sys	0.0s
Peak spill	0 B

branch

Metric	Value
Wall time	30.0s
Peak memory	12.5 MiB
Avg memory	7.8 MiB
CPU user	27.0s
CPU sys	0.0s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-06-09T17:59:44Z

🤖 Arrow criterion benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           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; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

group                                       main                                   variantbuilder-count-check
-----                                       ----                                   --------------------------
bench_extend_metadata_builder               1.00     21.4±0.09ms        ? ?/sec    1.00     21.4±0.06ms        ? ?/sec
bench_object_field_names_reverse_order      1.10     13.4±0.38ms        ? ?/sec    1.00     12.2±0.10ms        ? ?/sec
bench_object_list_partially_same_schema     1.00  1057.7±13.57µs        ? ?/sec    1.00  1054.3±14.47µs        ? ?/sec
bench_object_list_same_schema               1.00     22.6±0.10ms        ? ?/sec    1.00     22.7±0.45ms        ? ?/sec
bench_object_list_unknown_schema            1.00     11.2±0.04ms        ? ?/sec    1.00     11.1±0.04ms        ? ?/sec
bench_object_partially_same_schema          1.00      2.7±0.01ms        ? ?/sec    1.00      2.7±0.01ms        ? ?/sec
bench_object_same_schema                    1.00     32.4±0.08ms        ? ?/sec    1.07     34.6±0.05ms        ? ?/sec
bench_object_unknown_schema                 1.00     13.4±0.04ms        ? ?/sec    1.00     13.4±0.12ms        ? ?/sec
iteration/unvalidated_fallible_iteration    1.00      2.0±0.00ms        ? ?/sec    1.00      2.0±0.01ms        ? ?/sec
iteration/validated_iteration               1.02     39.0±0.48µs        ? ?/sec    1.00     38.1±0.73µs        ? ?/sec
validation/unvalidated_construction         1.02      4.4±0.04µs        ? ?/sec    1.00      4.4±0.03µs        ? ?/sec
validation/validated_construction           1.00     64.4±0.56µs        ? ?/sec    1.00     64.6±0.93µs        ? ?/sec
validation/validation_cost                  1.00     56.9±0.46µs        ? ?/sec    1.00     56.9±0.46µs        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	140.0s
Peak memory	53.9 MiB
Avg memory	16.5 MiB
CPU user	131.9s
CPU sys	1.8s
Peak spill	0 B

branch

Metric	Value
Wall time	135.0s
Peak memory	54.0 MiB
Avg memory	17.0 MiB
CPU user	129.8s
CPU sys	1.4s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-06-09T18:00:34Z

🤖 Arrow criterion benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           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; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

group                                                                                             main                                   variantbuilder-count-check
-----                                                                                             ----                                   --------------------------
batch_json_string_to_variant json_list 8k string                                                  1.00     22.1±0.23ms        ? ?/sec    1.01     22.5±0.19ms        ? ?/sec
batch_json_string_to_variant object - 1 depth(100 fields) random_json(2436 bytes per document)    1.00    153.7±0.78ms        ? ?/sec    1.00    153.3±0.45ms        ? ?/sec
batch_json_string_to_variant object - 1 depth(200 fields) random_json(4871 bytes per document)    1.01    324.5±1.43ms        ? ?/sec    1.00    321.4±1.03ms        ? ?/sec
batch_json_string_to_variant random_json(2633 bytes per document)                                 1.00    228.3±0.91ms        ? ?/sec    1.00    227.4±0.52ms        ? ?/sec
batch_json_string_to_variant repeated_struct 8k string                                            1.01      6.5±0.07ms        ? ?/sec    1.00      6.5±0.04ms        ? ?/sec
variant_get_primitive                                                                             1.00    634.8±5.34ns        ? ?/sec    1.00    632.8±4.16ns        ? ?/sec
variant_get_shredded_utf8                                                                         1.01    635.1±3.99ns        ? ?/sec    1.00    628.9±3.61ns        ? ?/sec

Resource Usage

base (merge-base)

Metric	Value
Wall time	160.0s
Peak memory	445.3 MiB
Avg memory	203.9 MiB
CPU user	151.6s
CPU sys	3.0s
Peak spill	0 B

branch

Metric	Value
Wall time	155.0s
Peak memory	443.7 MiB
Avg memory	211.8 MiB
CPU user	150.8s
CPU sys	2.4s
Peak spill	0 B

File an issue against this benchmark runner

sdf-jkl · 2026-06-09T18:26:16Z

everything looks fine other than -

bench_object_same_schema                    1.00     32.4±0.08ms        ? ?/sec    1.07     34.6±0.05ms        ? ?/sec

Add VariantBuilder values check

498c18a

github-actions Bot added the parquet-variant parquet-variant* crates label May 25, 2026

sdf-jkl added 6 commits May 26, 2026 12:03

Add try version of the APIs + cleanup

93c39cf

move try_new_list/try_new_object to VariantBuilder from `Varian…

168755f

…tBuidlerExt` trait

Merge branch 'main' of https://github.com/apache/arrow-rs into varian…

16a7fc1

…tbuilder-count-check

add inline for perf

ca42e23

fmt

b5aeaaf

clippy

f372e3a

sdf-jkl marked this pull request as ready for review May 28, 2026 22:09

alamb approved these changes Jun 3, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Variant] Add `VariantBuilder` values check#10016

[Variant] Add `VariantBuilder` values check#10016
sdf-jkl wants to merge 7 commits into
apache:mainfrom
sdf-jkl:variantbuilder-count-check

sdf-jkl commented May 25, 2026 •

edited

Loading

Uh oh!

sdf-jkl commented May 28, 2026

Uh oh!

alamb left a comment •

edited

Loading

Uh oh!

sdf-jkl commented Jun 9, 2026

Uh oh!

alamb commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

sdf-jkl commented Jun 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

sdf-jkl commented May 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

sdf-jkl commented May 28, 2026

Uh oh!

alamb left a comment • 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!

sdf-jkl commented Jun 9, 2026

Uh oh!

alamb commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

adriangbot commented Jun 9, 2026

Uh oh!

sdf-jkl commented Jun 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

sdf-jkl commented May 25, 2026 •

edited

Loading

alamb left a comment •

edited

Loading