Merge oracle coverage for schema objects; fix both-drop resurrection

[timsehn]

Fills the merge × schema-objects oracle gap (follow-on to the #1657 add-oracle coverage).

Coverage added (19 scenarios, live-Dolt comparison)

• Views: add ours/theirs, both-add identical (clean) and different (conflict), modify theirs, both-modify different (conflict), modify-vs-drop (conflict), both-drop, and in-transaction conflict inspection (dolt_conflicts count parity under BEGIN).
• Triggers: add-theirs fires post-merge, both-add identical fires once, both-add different (conflict), modify-vs-drop (conflict), both-drop stops firing. Trigger bodies can't share one script across dialects, so these use new oracle_dual_poststate/oracle_dual_error helpers with per-system setups and a shared post-state probe.
• Indexes: add-theirs, both-add same/different names, same-name different definition (conflict), and drop-vs-data-change (dual setup for DROP INDEX dialect).

Bug found and fixed

merge_view_both_drop and merge_trigger_both_drop failed on master: when both branches dropped a view or trigger, mergedSchemaChoice fell through to the ancestor entry and the merged catalog resurrected the dropped object. Dolt keeps it dropped. Fixed by returning no entry when both sides changed the name and neither side still has it — the same chokepoint serves hidden autoindex rows, so those can't resurrect either.

Tests

Both new both-drop scenarios fail before the fix and pass after. Full local regression green: battery 88/88 suites, C corpus 309,830/309,830, merge oracle 67/67, schema_merge 37/37, fk_merge 39/39, conflicts 9/9, index×VC matrix 38/38.

🤖 Generated with Claude Code

[github-actions]

Version-Control Performance Ceilings

Runs: median of 3 executions per benchmark, excluding fixture setup. The
working-set fixtures use 800 committed tables with 125 rows each
(100000 rows total); data-dirty cases update 1 row in 40
tables, and schema-dirty cases update 1 row plus add a column in 20 tables.
Branch tests use 300 branches, checkout uses a clean 400-table
branch switch over 100000 rows, and merge
tests use 100000-row tables with 2000 changed or conflicting
rows per side.

Benchmark	Median ms	Ceiling ms	Result
status_clean_many_tables	90	200	PASS
status_dirty_many_tables	93	200	PASS
diff_regular_working_one_table	85	150	PASS
diff_regular_working_many_tables	98	200	PASS
diff_stat_working_many_tables	98	200	PASS
diff_schema_working_many_tables	99	200	PASS
branch_list_many_branches	24	100	PASS
branch_create_delete	26	100	PASS
checkout_branch_clean	58	200	PASS
merge_data_no_conflicts	31	150	PASS
merge_schema_no_conflicts	23	100	PASS
merge_data_conflicts	129	250	PASS
merge_data_conflicts_with_resolve	128	250	PASS

[github-actions]

Sysbench-Style Benchmark (composite PK): Doltlite vs SQLite

Companion to the classic Sysbench-Style Benchmark. Every workload here
runs against tables with a 2-column INTEGER PRIMARY KEY(a, b) WITHOUT ROWID.
Individual ratios gated at 2.5×; section averages gated at 2×. Autocommit writes use 6× / 5×.

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	33,010	40,560	1.23
oltp_range_select	18,488	21,525	1.16
oltp_sum_range	17,763	20,700	1.17
oltp_order_range	3,528	3,863	1.09
oltp_distinct_range	4,633	4,837	1.04
oltp_index_scan	4,727	6,132	1.30
select_random_points	28,003	32,066	1.15
select_random_ranges	7,720	9,071	1.18
covering_index_scan	4,171	4,410	1.06
groupby_scan	36,148	38,651	1.07
index_join	8,203	10,690	1.30
index_join_scan	4,051	5,548	1.37
types_table_scan	1,125,593	1,424,732	1.27
table_scan	1,395,621	1,574,033	1.13
oltp_read_only	154,866	173,129	1.12
Average			1.18

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	247,124	367,030	1.49
oltp_insert	19,311	37,554	1.94
oltp_update_index	71,055	124,337	1.75
oltp_update_non_index	52,682	86,246	1.64
oltp_delete_insert	51,932	100,954	1.94
oltp_write_only	28,079	60,144	2.14
types_delete_insert	33,413	57,639	1.73
oltp_read_write	108,169	163,527	1.51
Average			1.77

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	103,880	65,751	0.63
oltp_range_select	26,406	24,392	0.92
oltp_sum_range	25,695	23,727	0.92
oltp_order_range	4,292	4,220	0.98
oltp_distinct_range	5,497	5,228	0.95
oltp_index_scan	11,978	9,075	0.76
select_random_points	38,683	37,757	0.98
select_random_ranges	15,137	11,869	0.78
covering_index_scan	11,447	7,082	0.62
groupby_scan	36,994	38,861	1.05
index_join	12,080	12,823	1.06
index_join_scan	5,015	5,923	1.18
types_table_scan	1,063,876	1,395,511	1.31
table_scan	1,323,121	1,550,520	1.17
oltp_read_only	258,048	209,693	0.81
Average			0.94

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	259,012	380,236	1.47
oltp_insert	25,441	46,320	1.82
oltp_update_index	94,919	142,127	1.50
oltp_update_non_index	75,751	105,429	1.39
oltp_delete_insert	74,757	119,261	1.60
oltp_write_only	50,437	79,014	1.57
types_delete_insert	50,623	69,561	1.37
oltp_read_write	131,974	182,062	1.38
Average			1.51

File-Backed (autocommit)

Each statement runs as its own transaction — exposes per-commit
fixed costs that the wrapped-in-BEGIN/COMMIT tests amortize away.
SQLite uses WAL mode with synchronous=FULL in this section so
the comparison uses SQLite's durable WAL autocommit path.

Reads

Reads have no commit cost; these are the same SQL files as the
File-Backed Reads section, included here for symmetry and to
catch any per-statement overhead doltlite pays on the read path.

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	58,160	65,595	1.13
oltp_range_select	21,111	24,429	1.16
oltp_sum_range	20,595	23,695	1.15
oltp_order_range	3,885	4,238	1.09
oltp_distinct_range	5,005	5,265	1.05
oltp_index_scan	7,409	9,018	1.22
select_random_points	32,754	37,725	1.15
select_random_ranges	10,473	11,802	1.13
covering_index_scan	6,881	7,200	1.05
groupby_scan	36,562	38,501	1.05
index_join	9,896	12,916	1.31
index_join_scan	4,520	5,848	1.29
types_table_scan	1,041,885	1,390,519	1.33
table_scan	1,229,979	1,531,424	1.25
oltp_read_only	180,399	206,950	1.15
Average			1.17

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	24,180	82,408	3.41
oltp_insert_ac	25,483	101,404	3.98
oltp_update_index_ac	27,473	113,059	4.12
oltp_update_non_index_ac	24,505	93,344	3.81
oltp_delete_insert_ac	29,009	104,192	3.59
oltp_write_only_ac	26,624	103,805	3.90
types_delete_insert_ac	21,730	90,599	4.17
oltp_read_write_ac	31,187	108,559	3.48
Average			3.81

100000 rows, median of 5 invocations per test; autocommit writes use 9, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (2.5x individual, 2x average; autocommit writes: 6x / 5x)

All tests within ceilings.

[github-actions]

Sysbench-Style Benchmark (BLOB PK): Doltlite vs SQLite

Companion to the classic Sysbench-Style Benchmark. Every workload here
runs against tables with a 16-byte big-endian BLOB PRIMARY KEY.
Individual ratios gated at 2.5×; section averages gated at 2×. Autocommit writes use 6× / 5×.

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	29,959	35,245	1.18
oltp_range_select	13,427	13,485	1.00
oltp_sum_range	11,725	12,762	1.09
oltp_order_range	3,097	3,206	1.04
oltp_distinct_range	4,153	4,287	1.03
oltp_index_scan	4,486	5,875	1.31
select_random_points	17,497	19,700	1.13
select_random_ranges	4,159	5,267	1.27
covering_index_scan	4,519	4,385	0.97
groupby_scan	33,877	34,809	1.03
index_join	6,861	8,915	1.30
index_join_scan	4,348	5,643	1.30
types_table_scan	1,127,983	1,265,358	1.12
table_scan	1,302,865	1,371,146	1.05
oltp_read_only	118,778	129,616	1.09
Average			1.13

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	240,554	336,533	1.40
oltp_insert	20,476	39,225	1.92
oltp_update_index	68,433	129,745	1.90
oltp_update_non_index	49,935	83,926	1.68
oltp_delete_insert	49,217	102,393	2.08
oltp_write_only	28,180	62,870	2.23
types_delete_insert	32,746	52,429	1.60
oltp_read_write	80,476	131,827	1.64
Average			1.81

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	112,904	63,885	0.57
oltp_range_select	21,335	16,518	0.77
oltp_sum_range	19,847	16,077	0.81
oltp_order_range	3,859	3,550	0.92
oltp_distinct_range	4,869	4,553	0.94
oltp_index_scan	12,754	9,097	0.71
select_random_points	25,454	23,271	0.91
select_random_ranges	12,588	8,158	0.65
covering_index_scan	13,036	7,733	0.59
groupby_scan	34,444	35,404	1.03
index_join	11,366	11,389	1.00
index_join_scan	5,164	6,162	1.19
types_table_scan	1,128,166	1,248,645	1.11
table_scan	1,316,059	1,361,452	1.03
oltp_read_only	238,342	170,207	0.71
Average			0.86

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	258,469	359,501	1.39
oltp_insert	30,682	52,670	1.72
oltp_update_index	103,749	165,486	1.60
oltp_update_non_index	75,794	109,885	1.45
oltp_delete_insert	80,382	133,313	1.66
oltp_write_only	53,336	87,443	1.64
types_delete_insert	51,127	71,587	1.40
oltp_read_write	110,592	156,693	1.42
Average			1.53

File-Backed (autocommit)

Each statement runs as its own transaction — exposes per-commit
fixed costs that the wrapped-in-BEGIN/COMMIT tests amortize away.
SQLite uses WAL mode with synchronous=FULL in this section so
the comparison uses SQLite's durable WAL autocommit path.

Reads

Reads have no commit cost; these are the same SQL files as the
File-Backed Reads section, included here for symmetry and to
catch any per-statement overhead doltlite pays on the read path.

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	58,656	63,934	1.09
oltp_range_select	16,915	16,706	0.99
oltp_sum_range	15,648	16,126	1.03
oltp_order_range	3,592	3,561	0.99
oltp_distinct_range	4,485	4,554	1.02
oltp_index_scan	7,659	9,317	1.22
select_random_points	20,774	23,463	1.13
select_random_ranges	7,144	8,359	1.17
covering_index_scan	8,205	7,734	0.94
groupby_scan	34,232	35,321	1.03
index_join	8,844	11,489	1.30
index_join_scan	4,919	6,356	1.29
types_table_scan	1,135,359	1,256,269	1.11
table_scan	1,328,374	1,380,715	1.04
oltp_read_only	158,682	172,171	1.09
Average			1.10

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	16,022	64,231	4.01
oltp_insert_ac	17,975	84,100	4.68
oltp_update_index_ac	19,945	96,493	4.84
oltp_update_non_index_ac	16,774	74,914	4.47
oltp_delete_insert_ac	17,672	87,010	4.92
oltp_write_only_ac	18,399	86,793	4.72
types_delete_insert_ac	15,706	76,226	4.85
oltp_read_write_ac	23,921	92,311	3.86
Average			4.54

100000 rows, median of 5 invocations per test; autocommit writes use 9, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (2.5x individual, 2x average; autocommit writes: 6x / 5x)

All tests within ceilings.

[github-actions]

Sysbench-Style Benchmark: Doltlite vs SQLite

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	24,176	29,532	1.22
oltp_range_select	9,591	12,557	1.31
oltp_sum_range	9,219	12,030	1.30
oltp_order_range	2,478	2,890	1.17
oltp_distinct_range	3,538	3,858	1.09
oltp_index_scan	3,949	5,233	1.33
select_random_points	9,877	10,905	1.10
select_random_ranges	2,864	4,009	1.40
covering_index_scan	4,113	4,144	1.01
groupby_scan	29,322	32,179	1.10
index_join	5,934	7,933	1.34
index_join_scan	3,460	4,527	1.31
types_table_scan	1,050,687	1,395,560	1.33
table_scan	1,219,407	1,514,228	1.24
oltp_read_only	102,887	120,403	1.17
Average			1.23

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	180,973	253,104	1.40
oltp_insert	15,714	29,688	1.89
oltp_update_index	51,418	95,046	1.85
oltp_update_non_index	36,486	61,506	1.69
oltp_delete_insert	45,512	71,049	1.56
oltp_write_only	22,323	44,994	2.02
types_delete_insert	25,029	40,246	1.61
oltp_read_write	68,529	110,216	1.61
Average			1.70

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	92,871	54,105	0.58
oltp_range_select	17,784	15,090	0.85
oltp_sum_range	16,182	14,741	0.91
oltp_order_range	3,294	3,175	0.96
oltp_distinct_range	4,367	4,211	0.96
oltp_index_scan	10,790	8,085	0.75
select_random_points	17,153	13,663	0.80
select_random_ranges	9,812	6,533	0.67
covering_index_scan	11,160	7,028	0.63
groupby_scan	30,250	32,570	1.08
index_join	10,351	10,058	0.97
index_join_scan	4,558	5,134	1.13
types_table_scan	1,103,050	1,407,111	1.28
table_scan	1,222,699	1,519,291	1.24
oltp_read_only	199,224	153,634	0.77
Average			0.91

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	199,585	274,340	1.37
oltp_insert	21,855	37,255	1.70
oltp_update_index	83,910	118,904	1.42
oltp_update_non_index	59,322	83,200	1.40
oltp_delete_insert	67,763	95,074	1.40
oltp_write_only	43,876	68,043	1.55
types_delete_insert	41,568	54,173	1.30
oltp_read_write	91,156	134,194	1.47
Average			1.45

File-Backed (autocommit)

Each statement runs as its own transaction — exposes per-commit
fixed costs that the wrapped-in-BEGIN/COMMIT tests amortize away.
SQLite uses WAL mode with synchronous=FULL in this section so
the comparison uses SQLite's durable WAL autocommit path.

Reads

Reads have no commit cost; these are the same SQL files as the
File-Backed Reads section, included here for symmetry and to
catch any per-statement overhead doltlite pays on the read path.

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	47,668	53,930	1.13
oltp_range_select	14,025	16,208	1.16
oltp_sum_range	13,602	15,129	1.11
oltp_order_range	3,802	3,646	0.96
oltp_distinct_range	4,428	4,409	1.00
oltp_index_scan	6,674	8,458	1.27
select_random_points	14,491	14,202	0.98
select_random_ranges	5,784	6,753	1.17
covering_index_scan	7,695	7,118	0.93
groupby_scan	32,346	34,353	1.06
index_join	8,085	10,333	1.28
index_join_scan	4,275	5,399	1.26
types_table_scan	1,073,373	1,388,363	1.29
table_scan	1,186,112	1,518,153	1.28
oltp_read_only	142,579	157,884	1.11
Average			1.13

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	21,409	80,895	3.78
oltp_insert_ac	25,129	108,712	4.33
oltp_update_index_ac	26,117	116,286	4.45
oltp_update_non_index_ac	24,348	88,837	3.65
oltp_delete_insert_ac	25,256	100,412	3.98
oltp_write_only_ac	23,523	97,405	4.14
types_delete_insert_ac	22,769	93,106	4.09
oltp_read_write_ac	28,077	110,178	3.92
Average			4.04

100000 rows, median of 5 invocations per test; autocommit writes use 9, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (2.5x individual, 2x average; autocommit writes: 6x / 5x)

All tests within ceilings.

[github-actions]

Sysbench-Style Benchmark (TEXT PK): Doltlite vs SQLite

Companion to the classic Sysbench-Style Benchmark. Every workload here
runs against tables with a 32-char hex TEXT PRIMARY KEY (UUID-shaped).
Individual ratios gated at 2.5×; section averages gated at 2×. Autocommit writes use 6× / 5×.

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	31,569	38,324	1.21
oltp_range_select	13,073	14,622	1.12
oltp_sum_range	12,719	14,370	1.13
oltp_order_range	3,084	3,151	1.02
oltp_distinct_range	4,048	4,171	1.03
oltp_index_scan	4,557	6,401	1.40
select_random_points	19,729	21,863	1.11
select_random_ranges	4,211	5,428	1.29
covering_index_scan	5,007	4,913	0.98
groupby_scan	32,214	33,564	1.04
index_join	6,907	9,377	1.36
index_join_scan	4,423	5,297	1.20
types_table_scan	1,109,464	1,388,073	1.25
table_scan	1,410,155	1,539,100	1.09
oltp_read_only	129,209	140,389	1.09
Average			1.15

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	234,046	361,419	1.54
oltp_insert	23,286	40,634	1.74
oltp_update_index	78,692	143,167	1.82
oltp_update_non_index	50,646	90,118	1.78
oltp_delete_insert	54,698	107,644	1.97
oltp_write_only	30,333	64,033	2.11
types_delete_insert	34,105	57,290	1.68
oltp_read_write	95,284	145,881	1.53
Average			1.77

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	101,520	63,453	0.63
oltp_range_select	21,406	17,376	0.81
oltp_sum_range	20,877	16,816	0.81
oltp_order_range	4,027	3,558	0.88
oltp_distinct_range	4,973	4,549	0.91
oltp_index_scan	12,134	9,158	0.75
select_random_points	28,957	25,579	0.88
select_random_ranges	11,423	7,841	0.69
covering_index_scan	13,350	7,498	0.56
groupby_scan	33,455	34,259	1.02
index_join	11,976	11,409	0.95
index_join_scan	5,650	5,832	1.03
types_table_scan	1,253,751	1,417,269	1.13
table_scan	1,520,502	1,549,853	1.02
oltp_read_only	236,677	177,338	0.75
Average			0.86

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	249,326	395,415	1.59
oltp_insert	43,311	53,265	1.23
oltp_update_index	119,166	175,739	1.47
oltp_update_non_index	90,888	114,058	1.25
oltp_delete_insert	91,761	135,634	1.48
oltp_write_only	98,671	85,421	0.87
types_delete_insert	56,031	76,555	1.37
oltp_read_write	145,409	165,809	1.14
Average			1.30

File-Backed (autocommit)

Each statement runs as its own transaction — exposes per-commit
fixed costs that the wrapped-in-BEGIN/COMMIT tests amortize away.
SQLite uses WAL mode with synchronous=FULL in this section so
the comparison uses SQLite's durable WAL autocommit path.

Reads

Reads have no commit cost; these are the same SQL files as the
File-Backed Reads section, included here for symmetry and to
catch any per-statement overhead doltlite pays on the read path.

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	55,172	62,694	1.14
oltp_range_select	17,098	17,235	1.01
oltp_sum_range	15,342	17,002	1.11
oltp_order_range	3,508	3,469	0.99
oltp_distinct_range	4,507	4,657	1.03
oltp_index_scan	7,575	9,160	1.21
select_random_points	21,808	25,869	1.19
select_random_ranges	6,903	7,861	1.14
covering_index_scan	8,407	7,338	0.87
groupby_scan	32,999	34,187	1.04
index_join	10,204	11,408	1.12
index_join_scan	5,151	6,077	1.18
types_table_scan	1,248,830	1,433,303	1.15
table_scan	1,541,438	1,558,950	1.01
oltp_read_only	168,506	179,687	1.07
Average			1.08

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	20,985	75,952	3.62
oltp_insert_ac	24,653	92,069	3.73
oltp_update_index_ac	26,856	110,221	4.10
oltp_update_non_index_ac	21,771	89,951	4.13
oltp_delete_insert_ac	24,665	101,329	4.11
oltp_write_only_ac	23,638	103,093	4.36
types_delete_insert_ac	21,677	92,139	4.25
oltp_read_write_ac	29,393	106,661	3.63
Average			3.99

100000 rows, median of 5 invocations per test; autocommit writes use 9, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (2.5x individual, 2x average; autocommit writes: 6x / 5x)

All tests within ceilings.