[fix](load) Fix Arrow stream load with uppercase columns

[sollhui]

What problem does this PR solve?

Issue Number: None

Related PR: None

Problem Summary: Arrow Stream Load through the Nereids load planner lowercased source column names even though Arrow schema field names are case-sensitive. For tables with uppercase target columns, the Arrow scan-slot path could resolve the source name through a case-sensitive type map and fail planning with a null type or construct a scan slot whose name no longer matched the Arrow field. This change preserves source column names for Arrow load format, including non-lowercase Arrow format spellings, and resolves Arrow scan slot types through the target table's case-insensitive column lookup.

Release note

Fix Arrow Stream Load for tables with uppercase column names.

Check List (For Author)

• Test: Checkstyle; Regression test added; Unit test attempted
  • mvn -pl fe-core -DskipTests checkstyle:check
  • ./run-fe-ut.sh --run org.apache.doris.nereids.load.NereidsLoadScanProviderTest (not completed locally due existing fe-catalog compilation errors in Column.java/ScalarType before reaching fe-core tests)
  • ./run-regression-test.sh --run -d load_p0/stream_load -s test_arrow_stream_load_uppercase_column -genOut (not completed locally because no Doris FE was listening on 127.0.0.1:9030)
• Behavior changed: Yes. Arrow load source column names are preserved in the Nereids load planner, and Arrow target column types are resolved case-insensitively.
• Does this need documentation: No

[sollhui]

run buildall

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[sollhui]

/review

[sollhui]

/review

[sollhui]

run buildall

[github-actions]

I found two issues that should be fixed before merge.

Critical checkpoints: the PR's goal is clear and mostly focused on preserving Arrow source-column case in the Nereids load planner, but the implementation does not cover accepted non-lowercase FORMAT AS spellings. No concurrency, lifecycle, persistence, FE/BE protocol, or config-change concerns were introduced by this small FE/load-planning change. I checked the related Nereids stream/broker load flow and the BE Arrow reader name check. Test coverage was added, but the new deterministic regression result should use generated qt_/order_qt_ output instead of direct assertions. I could not run the FE unit or regression tests in this checkout because thirdparty/installed and thirdparty/installed/bin/protoc are missing.

Subagent conclusions: optimizer-rewrite produced MAIN-001 and tests-session-config produced MAIN-002; both were independently verified and accepted for inline comments. Convergence round 1 ended with both subagents replying NO_NEW_VALUABLE_FINDINGS for this final comment set.

User focus: no additional user-provided review focus was present.

[github-actions]

Additional changes requested after refreshing the review to head bfa4f6a.

I found two distinct correctness issues that are not covered by the existing inline comments: broker LOAD still re-lowercases Arrow source names through the legacy DataDescription conversion, and fixed partial-update metadata does not use the same case-insensitive column resolution that the new Arrow slot typing now uses.

Critical checkpoints: the PR goal is clear and the change is narrow, but parallel load paths and write-path metadata are incomplete. No new concurrency, lifecycle, config, persistence, or mixed-version protocol surface was introduced by the edited FE planner code. The main risk is data/write correctness for Arrow load planning, especially broker LOAD and MoW fixed partial update. Test coverage was added for stream load, but it does not cover these two paths. I could not run FE unit or regression tests in this checkout because thirdparty/installed and thirdparty/installed/bin/protoc are missing; static diff hygiene via git diff --check passed.

Duplicate-preflight: existing GitHub comments already cover the case-sensitive FORMAT AS ARROW spelling and the regression qt_/order_qt_ output issue, so I did not repeat those.

Subagent conclusions: OR-1 was merged into the broker legacy-conversion inline comment. TSC-1 was dismissed as stale/out of the live PR scope. After the PR head moved, convergence round final-2 ended with both optimizer-rewrite and tests-session-config replying NO_NEW_VALUABLE_FINDINGS for the current ledger/comment set.

User focus: no additional user-provided review focus was present.

[github-actions]

Codex automated review failed and did not complete.

Error: Codex completed, but no new pull request review was submitted for the current head SHA.
Workflow run: https://github.com/apache/doris/actions/runs/28527766728

Please inspect the workflow logs and rerun the review after the underlying issue is resolved.

[sollhui]

/review

[sollhui]

run buildall

[github-actions]

Automated review follow-up for current head e08f8130e5a6df1c2c1aaa932f174f0c56dca1f7.

I found no new valuable inline comments to add without duplicating existing review threads. The current head appears to have addressed the earlier non-lowercase Arrow format spelling issue and the regression-result-framework issue. The two remaining substantiated concerns are already filed as inline threads:

• 3507270606: broker LOAD still passes preserved Arrow names through legacy DataDescription, which lowercases them before converting back to NereidsBrokerFileGroup.
• 3507270613: fixed partial-update metadata still collects input columns with exact-case matching while Arrow scan-slot type resolution now accepts case-insensitive source names.

Critical checkpoint conclusions:

• Goal and proof: the PR targets Arrow stream load with uppercase column names; direct stream-load paths are covered by the new regression and FE unit coverage, but broker LOAD and fixed partial-update variants remain covered by existing unresolved review threads.
• Scope: implementation is narrow in the Nereids load description/scan-provider path plus focused tests.
• Concurrency/lifecycle: no new concurrency, locking, or lifecycle-sensitive state is introduced.
• Configuration/session/compatibility: no new config items or serialized formats are added; the relevant compatibility risk is the existing legacy broker conversion path.
• Parallel paths: stream load was updated, but broker load and fixed partial-update metadata remain the important parallel paths, already called out inline.
• Tests/results: the current regression uses order_qt_ with an .out file; I could not run FE or regression tests in this checkout because thirdparty/installed and thirdparty/installed/bin/protoc are absent.

Subagent conclusions: optimizer-rewrite and tests-session-config both completed the final convergence round with NO_NEW_VALUABLE_FINDINGS for the same final ledger/comment set. No subagent candidates became new inline comments; the only live findings were merged as duplicates of existing GitHub threads 3507270606 and 3507270613.

User focus: no additional user-provided review focus was present.

[hello-stephen]

TPC-H: Total hot run time: 29684 ms

machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
scripts: https://github.com/apache/doris/tree/master/tools/tpch-tools
Tpch sf100 test result on commit e08f8130e5a6df1c2c1aaa932f174f0c56dca1f7, data reload: false
------ Round 1 ----------------------------------
============================================
q1	17819	4081	4023	4023
q2	2081	324	195	195
q3	10312	1420	827	827
q4	4692	469	350	350
q5	7531	860	585	585
q6	176	168	140	140
q7	761	867	619	619
q8	9360	1644	1592	1592
q9	5712	4414	4387	4387
q10	6718	1780	1532	1532
q11	499	343	330	330
q12	735	554	441	441
q13	18117	3424	2772	2772
q14	269	266	235	235
q15	q16	789	776	710	710
q17	1059	1035	939	939
q18	6962	5708	5517	5517
q19	1326	1206	1123	1123
q20	757	665	595	595
q21	5945	2589	2474	2474
q22	434	359	298	298
Total cold run time: 102054 ms
Total hot run time: 29684 ms
----- Round 2, with runtime_filter_mode=off -----
============================================
q1	4324	4274	4296	4274
q2	288	309	213	213
q3	4623	5012	4417	4417
q4	2053	2157	1363	1363
q5	4487	4325	4343	4325
q6	233	172	125	125
q7	1740	2042	1831	1831
q8	2524	2210	2189	2189
q9	8072	8102	7803	7803
q10	4719	4682	4221	4221
q11	652	417	399	399
q12	767	770	546	546
q13	3331	3597	2902	2902
q14	292	299	284	284
q15	q16	698	721	637	637
q17	1345	1341	1330	1330
q18	7929	7282	7299	7282
q19	1157	1147	1079	1079
q20	2207	2215	1928	1928
q21	5304	4616	4436	4436
q22	517	452	411	411
Total cold run time: 57262 ms
Total hot run time: 51995 ms

[hello-stephen]

TPC-DS: Total hot run time: 173665 ms

machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
scripts: https://github.com/apache/doris/tree/master/tools/tpcds-tools
TPC-DS sf100 test result on commit e08f8130e5a6df1c2c1aaa932f174f0c56dca1f7, data reload: false
query5	4300	639	480	480
query6	479	223	200	200
query7	4844	560	344	344
query8	343	191	169	169
query9	8784	4054	4062	4054
query10	461	367	336	336
query11	5873	2387	2156	2156
query12	165	106	101	101
query13	1252	608	415	415
query14	6276	5327	4984	4984
query14_1	4331	4309	4322	4309
query15	215	204	178	178
query16	1039	498	470	470
query17	1172	733	597	597
query18	2470	472	363	363
query19	214	194	160	160
query20	121	112	107	107
query21	240	151	135	135
query22	13607	13505	13475	13475
query23	17325	16564	16205	16205
query23_1	16312	16358	16362	16358
query24	7507	1781	1318	1318
query24_1	1321	1333	1287	1287
query25	584	458	392	392
query26	1326	331	206	206
query27	2613	619	373	373
query28	4409	2039	2025	2025
query29	1124	643	499	499
query30	346	266	231	231
query31	1117	1105	990	990
query32	135	63	62	62
query33	527	325	264	264
query34	1183	1171	643	643
query35	772	804	663	663
query36	1385	1419	1243	1243
query37	159	112	94	94
query38	1921	1716	1628	1628
query39	940	906	895	895
query39_1	889	888	876	876
query40	245	162	149	149
query41	66	62	61	61
query42	93	99	93	93
query43	320	321	275	275
query44	1440	763	764	763
query45	197	189	181	181
query46	1080	1179	760	760
query47	2391	2339	2267	2267
query48	406	427	305	305
query49	588	441	317	317
query50	1088	432	343	343
query51	4373	4416	4321	4321
query52	86	86	73	73
query53	263	274	196	196
query54	276	241	222	222
query55	74	70	65	65
query56	310	282	315	282
query57	1473	1435	1328	1328
query58	303	249	272	249
query59	1591	1643	1422	1422
query60	304	272	248	248
query61	155	149	149	149
query62	691	651	579	579
query63	252	210	207	207
query64	2514	752	613	613
query65	4860	4785	4764	4764
query66	1817	520	384	384
query67	29626	29637	29383	29383
query68	3452	1608	932	932
query69	398	310	268	268
query70	1092	995	963	963
query71	371	331	324	324
query72	3006	2667	2340	2340
query73	808	773	424	424
query74	5090	4993	4769	4769
query75	2617	2596	2226	2226
query76	2332	1199	775	775
query77	355	378	291	291
query78	12512	12568	11910	11910
query79	1474	1112	763	763
query80	1292	550	465	465
query81	546	333	280	280
query82	990	153	123	123
query83	390	322	286	286
query84	286	162	129	129
query85	987	598	501	501
query86	469	292	274	274
query87	1832	1821	1770	1770
query88	3718	2786	2756	2756
query89	460	411	359	359
query90	1971	192	195	192
query91	202	184	160	160
query92	63	61	56	56
query93	1688	1532	984	984
query94	776	356	326	326
query95	768	559	469	469
query96	1016	807	369	369
query97	2687	2742	2571	2571
query98	220	206	203	203
query99	1165	1141	1041	1041
Total cold run time: 260485 ms
Total hot run time: 173665 ms

[hello-stephen]

ClickBench: Total hot run time: 25.3 s

machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
scripts: https://github.com/apache/doris/tree/master/tools/clickbench-tools
ClickBench test result on commit e08f8130e5a6df1c2c1aaa932f174f0c56dca1f7, data reload: false
query1	0.01	0.01	0.01
query2	0.10	0.05	0.05
query3	0.26	0.14	0.14
query4	1.61	0.14	0.13
query5	0.23	0.22	0.22
query6	1.24	1.11	1.08
query7	0.04	0.01	0.01
query8	0.06	0.04	0.04
query9	0.38	0.33	0.32
query10	0.57	0.56	0.56
query11	0.20	0.13	0.14
query12	0.18	0.15	0.14
query13	0.48	0.48	0.47
query14	1.03	1.00	1.01
query15	0.61	0.59	0.60
query16	0.34	0.34	0.33
query17	1.10	1.12	1.11
query18	0.23	0.21	0.21
query19	2.01	1.86	1.93
query20	0.01	0.01	0.01
query21	15.44	0.21	0.14
query22	4.85	0.05	0.05
query23	16.17	0.31	0.12
query24	2.97	0.42	0.32
query25	0.10	0.05	0.05
query26	0.76	0.22	0.14
query27	0.04	0.05	0.04
query28	3.51	0.93	0.53
query29	12.54	4.43	3.52
query30	0.27	0.14	0.15
query31	2.76	0.60	0.32
query32	3.22	0.60	0.49
query33	3.33	3.21	3.27
query34	15.80	4.22	3.50
query35	3.57	3.53	3.55
query36	0.57	0.46	0.43
query37	0.09	0.06	0.06
query38	0.06	0.04	0.03
query39	0.04	0.02	0.02
query40	0.18	0.16	0.16
query41	0.08	0.04	0.03
query42	0.03	0.03	0.03
query43	0.04	0.04	0.03
Total cold run time: 97.11 s
Total hot run time: 25.3 s

[hello-stephen]

FE Regression Coverage Report

Increment line coverage 11.69% (9/77) 🎉
Increment coverage report
Complete coverage report

[morrySnow]

Code Review — #65127 Fix Arrow stream load with uppercase columns

The fix is functionally correct for the Arrow use case and well-tested. However, the same class of bug exists for the NATIVE format path but was left unfixed, and the approach adds yet another format-specific special case to an already-fragmented codebase. See inline comments for details.

[sollhui]

/review

[sollhui]

run buildall

[github-actions]

Automated review for PR 65127 at d930216.

I found one remaining correctness issue and requested changes inline.

Critical checkpoint conclusions:

• Goal/test proof: the PR addresses Arrow source-column case preservation, legacy broker conversion, and Native scan-slot type lookup, with regression and FE unit coverage added. The broker fixed partial-update path still has a correctness gap.
• Scope: the change is mostly focused on FE load/Nereids case handling and the paired tests.
• Concurrency/lifecycle/config/persistence: no new locks, long-lived state, configs, or persisted formats are introduced.
• Parallel paths/compatibility: a parallel broker-load fixed partial-update collector still uses exact-case matching before the new case-insensitive scan-slot resolution.
• Tests: the regression suite now uses order_qt output and has a committed .out file; unit coverage exists for Arrow and Native scan-slot type resolution, but broker Arrow fixed partial update with mixed-case target columns is not covered.
• Validation limits: this checkout is missing .worktree_initialized and thirdparty/installed/bin/protoc, so I did static/GitHub-metadata review only and did not run FE or regression tests.

Subagent conclusions:

• optimizer-rewrite proposed OR-01; I accepted it as M-OR-01 and submitted the broker-load partial-update inline comment.
• tests-session-config proposed TS-1 as a duplicate of existing thread discussion_r3507270613; I suppressed it as already covered.
• Convergence round 1 ended with both live subagents reporting NO_NEW_VALUABLE_FINDINGS for the final ledger/comment set.

User focus: no additional user-provided review focus was present.

[sollhui]

/review

[sollhui]

run buildall

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

I reviewed the current head db15a1d3052bdd1dc8aceecbb50aa96f0c991169 against the GitHub PR file list and existing review threads. I did not find a new distinct issue that should be added as another inline comment.

Critical checkpoint conclusions:

• Goal and proof: the current patch preserves Arrow/JSON file-field case where the reader needs exact source names, keeps columnsFromPath canonicalized, resolves Arrow/Native scan-slot types through the table column lookup, and applies case-insensitive matching for fixed partial-update input columns. Unit and regression coverage was added for the main Arrow uppercase paths and the path-column canonicalization case.
• Scope: the change is focused on load column-name case handling across Nereids, legacy conversion, stream load, broker load, partial update metadata, and the related regression/unit coverage.
• Concurrency/lifecycle: no new concurrency, lock ordering, async lifecycle, or static initialization behavior was introduced in the reviewed paths.
• Config/session compatibility: no new config item was added. Format-name behavior is centralized in Util.isCasePreservingFormat, and the non-lowercase Arrow format spelling case is covered.
• Parallel paths: I checked both Nereids and legacy DataDescription, stream-load and broker-load partial-update collection, and the BE Arrow/Native reader interactions. The earlier review threads for Arrow spelling, legacy broker relowering, Native null type, partial update, columns-from-path, and result-framework output are addressed in the current patch or already covered by existing comments.
• Tests: the new regression uses order_qt_* with a committed .out, and the FE unit test covers explicit Arrow source case, non-lowercase Arrow format spelling, path-column lowercasing, Arrow/Native type lookup, and import-column case-insensitive matching. I did not rerun FE or regression tests in this review; the local checkout is grafted and lacks the prompt base object, so I used GitHub PR metadata/patch as the authoritative scope. A simple scan of added patch lines found no trailing whitespace.
• Persistence/transaction/write behavior: no EditLog/storage-format change was introduced. The fixed partial-update paths now forward canonical table column names after case-insensitive matching, which preserves the downstream BE exact-name partial-update contract.
• Observability/performance: no new observability requirement or obvious hot-path performance issue stood out; the added checks are small planner-time lookups.

User focus: no additional user-provided review focus was supplied.

Subagent conclusions: the optimizer-rewrite and tests-session-config subagents found no new candidate findings in their initial passes. After the main ledger recorded the duplicate/dismissed points and the proposed comment-only final set, convergence round 1 ended with both live subagents replying NO_NEW_VALUABLE_FINDINGS for the same ledger/comment set.

[hello-stephen]

TPC-H: Total hot run time: 29754 ms

machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
scripts: https://github.com/apache/doris/tree/master/tools/tpch-tools
Tpch sf100 test result on commit db15a1d3052bdd1dc8aceecbb50aa96f0c991169, data reload: false
------ Round 1 ----------------------------------
============================================
q1	17978	4128	4141	4128
q2	2026	321	201	201
q3	10309	1384	806	806
q4	4682	471	340	340
q5	7491	864	561	561
q6	179	167	138	138
q7	760	843	627	627
q8	9325	1623	1640	1623
q9	5622	4408	4392	4392
q10	6785	1805	1518	1518
q11	489	335	311	311
q12	694	547	434	434
q13	18098	3433	2806	2806
q14	272	260	252	252
q15	q16	791	778	711	711
q17	1024	1004	918	918
q18	6836	5721	5643	5643
q19	1330	1310	1076	1076
q20	812	646	556	556
q21	5912	2568	2412	2412
q22	428	360	301	301
Total cold run time: 101843 ms
Total hot run time: 29754 ms
----- Round 2, with runtime_filter_mode=off -----
============================================
q1	4333	4267	4277	4267
q2	303	327	224	224
q3	4614	4974	4391	4391
q4	2061	2144	1370	1370
q5	4439	4322	4303	4303
q6	236	177	130	130
q7	1751	1962	1939	1939
q8	2606	2258	2252	2252
q9	8136	8259	7815	7815
q10	4814	4805	4342	4342
q11	599	461	425	425
q12	770	772	579	579
q13	3322	3537	2927	2927
q14	299	308	283	283
q15	q16	708	741	645	645
q17	1348	1357	1328	1328
q18	8350	7526	7416	7416
q19	1101	1072	1037	1037
q20	2218	2212	1947	1947
q21	5234	4640	4496	4496
q22	524	463	404	404
Total cold run time: 57766 ms
Total hot run time: 52520 ms

[hello-stephen]

TPC-DS: Total hot run time: 180727 ms

machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
scripts: https://github.com/apache/doris/tree/master/tools/tpcds-tools
TPC-DS sf100 test result on commit db15a1d3052bdd1dc8aceecbb50aa96f0c991169, data reload: false
query5	4313	606	493	493
query6	455	210	202	202
query7	4835	598	333	333
query8	330	182	174	174
query9	8783	4046	4056	4046
query10	479	375	308	308
query11	5911	2356	2151	2151
query12	159	103	103	103
query13	1263	597	426	426
query14	6283	5298	4934	4934
query14_1	4282	4263	4279	4263
query15	211	212	177	177
query16	1013	464	449	449
query17	1140	734	590	590
query18	2460	470	350	350
query19	207	193	150	150
query20	112	110	109	109
query21	232	158	139	139
query22	13635	13684	13373	13373
query23	17420	16524	16276	16276
query23_1	16381	16371	16252	16252
query24	7465	1775	1296	1296
query24_1	1378	1331	1327	1327
query25	559	475	420	420
query26	1338	344	211	211
query27	2619	572	388	388
query28	4410	1993	2038	1993
query29	1124	655	505	505
query30	337	292	229	229
query31	1119	1090	1001	1001
query32	110	61	60	60
query33	535	324	257	257
query34	1168	1144	656	656
query35	770	786	668	668
query36	1374	1388	1227	1227
query37	151	112	90	90
query38	1874	1712	1646	1646
query39	957	915	882	882
query39_1	872	879	896	879
query40	240	155	133	133
query41	64	67	64	64
query42	93	89	91	89
query43	319	317	283	283
query44	1407	798	775	775
query45	204	184	173	173
query46	1075	1197	765	765
query47	2381	2336	2251	2251
query48	410	403	291	291
query49	589	413	330	330
query50	1004	434	345	345
query51	10991	10994	10766	10766
query52	84	85	73	73
query53	272	284	202	202
query54	274	234	211	211
query55	74	70	65	65
query56	282	304	285	285
query57	1450	1406	1353	1353
query58	281	244	235	235
query59	1624	1615	1410	1410
query60	293	269	244	244
query61	164	150	150	150
query62	688	650	582	582
query63	248	205	204	204
query64	2813	1012	894	894
query65	4890	4830	4817	4817
query66	1820	501	377	377
query67	29448	29408	29382	29382
query68	3310	1482	915	915
query69	403	296	265	265
query70	1071	980	959	959
query71	341	302	325	302
query72	3018	2665	2428	2428
query73	827	813	444	444
query74	5093	4981	4750	4750
query75	2605	2588	2225	2225
query76	2327	1163	776	776
query77	341	374	281	281
query78	12194	12307	11698	11698
query79	1410	1157	760	760
query80	1287	539	464	464
query81	550	325	280	280
query82	611	157	123	123
query83	367	318	288	288
query84	278	161	126	126
query85	959	578	490	490
query86	428	307	293	293
query87	1823	1821	1745	1745
query88	3705	2810	2820	2810
query89	452	406	347	347
query90	1930	204	188	188
query91	201	200	161	161
query92	63	60	58	58
query93	1770	1482	1062	1062
query94	722	364	320	320
query95	751	601	457	457
query96	1080	784	375	375
query97	2704	2694	2572	2572
query98	209	206	205	205
query99	1147	1183	1041	1041
Total cold run time: 265885 ms
Total hot run time: 180727 ms

[hello-stephen]

ClickBench: Total hot run time: 25.04 s

machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
scripts: https://github.com/apache/doris/tree/master/tools/clickbench-tools
ClickBench test result on commit db15a1d3052bdd1dc8aceecbb50aa96f0c991169, data reload: false
query1	0.01	0.01	0.00
query2	0.09	0.05	0.05
query3	0.25	0.16	0.15
query4	1.61	0.14	0.14
query5	0.24	0.23	0.22
query6	1.25	1.06	1.06
query7	0.03	0.01	0.00
query8	0.06	0.03	0.03
query9	0.37	0.31	0.32
query10	0.54	0.59	0.55
query11	0.19	0.14	0.14
query12	0.18	0.15	0.14
query13	0.47	0.46	0.48
query14	1.00	0.99	1.00
query15	0.62	0.59	0.59
query16	0.32	0.32	0.32
query17	1.15	1.07	1.10
query18	0.22	0.20	0.21
query19	2.05	1.96	1.99
query20	0.01	0.01	0.02
query21	15.45	0.21	0.14
query22	4.99	0.06	0.05
query23	16.12	0.31	0.12
query24	2.94	0.42	0.31
query25	0.12	0.05	0.04
query26	0.72	0.21	0.16
query27	0.05	0.04	0.03
query28	3.49	0.93	0.54
query29	12.49	4.06	3.24
query30	0.28	0.17	0.15
query31	2.78	0.60	0.31
query32	3.22	0.59	0.49
query33	3.30	3.28	3.22
query34	15.43	4.24	3.54
query35	3.55	3.52	3.55
query36	0.56	0.44	0.44
query37	0.09	0.07	0.06
query38	0.05	0.04	0.03
query39	0.04	0.02	0.02
query40	0.18	0.17	0.15
query41	0.09	0.04	0.03
query42	0.04	0.03	0.03
query43	0.05	0.03	0.04
Total cold run time: 96.69 s
Total hot run time: 25.04 s

fixed