fix(revert): match Dolt 2.0.5 dirty-set rules (PR dolt#11073 + follow-up) by timsehn · Pull Request #992 · dolthub/doltlite · GitHub

timsehn · 2026-05-20T22:26:26Z

Summary

Match the Dolt 2.0.5 revert semantics so the version-control oracle tests pass against the current Dolt release.

Two rules, mirroring git revert exactly:

Any staged change refuses the revert — even when the staged table is unrelated to the revert target. (Git's "index must be clean" requirement.)
Unstaged changes refuse only if they touch a table the revert would also touch — disjoint unstaged changes are allowed and get folded into the revert commit.

What was wrong

Before this PR, doltlite refused only when a dirty change overlapped with a table the revert would touch. That matched Dolt's behavior up through 2.0.4 but diverged from 2.0.5, which added the stricter index-clean rule on top.

Caught by vc_oracle_revert_test.sh's revert_head_staged_unrelated_table_committed case once CI upgraded to Dolt 2.0.5.

What changed

Replaced three internal helpers (doltliteRevertHasDirtyTouchedTables, doltliteCollectRevertTouchedTables, plus the name-list utility helpers) with a single doltliteRevertCheckDirty that does both checks inline:

If session-staged catalog hash != session-head hash, flag dirty regardless of which tables differ.
Else walk every table that differs between working and head; if the revert target catalog and parent catalog also differ on that same table, flag dirty.

Error message updated to match Dolt's exact wording:

Your local changes would be overwritten by revert.
hint: Please commit your changes before you revert.

Net diff: -131 lines, +58 lines (simpler, single helper).

References

Dolt PR dolthub/dolt#11073 — "dolt revert to more closely match git"
Dolt commit 014f1d8 — "align revert dirty-set rules with git" (the follow-up that added the index-clean rule)
Both are in Dolt v2.0.5.

Test plan

Verified locally against /tmp/dolt-2.0.5 (the binary downloaded from the v2.0.5 release):

vc_oracle_revert_test.sh — 6/6 pass (was 5/6 with revert_head_staged_unrelated_table_committed failing before this change)
vc_oracle_revert_cherrypick_test.sh — 45/45 pass
doltlite_merge.sh — 91/91 pass
doltlite_conflicts.sh — 40/40 pass
doltlite_advanced.sh — 140/140 pass

🤖 Generated with Claude Code

…-up) Dolt 2.0.5 tightened dolt_revert's dirty-set check to match git's behavior: 1. Any staged change refuses the revert, even when the staged table is unrelated to the revert target. Mirrors git's "index must be clean" requirement — otherwise the revert commit would mix staged work with the inverse-change commit. 2. Unstaged changes refuse the revert only if they overlap with a table the revert would touch. Disjoint unstaged changes are allowed and get folded into the revert commit. Previous doltlite behavior was "refuse only if dirty change touches a table the revert would also touch", which allowed staged-disjoint to proceed. That diverged from Dolt 2.0.5 on revert_head_staged_unrelated_table_committed (caught by vc_oracle_revert_test.sh once Dolt was upgraded to 2.0.5). Replace doltliteRevertHasDirtyTouchedTables / doltliteCollectRevertTouchedTables / their internal name-list helpers with a single doltliteRevertCheckDirty that does both checks inline: * If session-staged catalog hash differs from session-head hash, flag dirty regardless of which tables differ. * Else compare working catalog to head; for each differing table, check whether the revert target catalog and parent catalog differ on that same table — if so, flag dirty. Error message also updated to match Dolt's: "Your local changes would be overwritten by revert.\nhint: ..." Verified against /tmp/dolt-2.0.5 binary downloaded from the v2.0.5 release: * vc_oracle_revert_test.sh: 6/6 pass (was 5/6 with the staged-disjoint case failing before this change) * vc_oracle_revert_cherrypick_test.sh: 45/45 pass * doltlite_merge.sh: 91/91 pass * doltlite_conflicts.sh: 40/40 pass * doltlite_advanced.sh: 140/140 pass References: Dolt PR dolthub/dolt#11073 ("dolt revert to more closely match git") and follow-up commit 014f1d8 ("align revert dirty-set rules with git") in dolthub/dolt v2.0.5. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

github-actions · 2026-05-20T22:33:25Z

Sysbench-Style Benchmark: Doltlite vs SQLite

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	21,127	27,682	1.31
oltp_range_select	8,452	12,819	1.52
oltp_sum_range	7,670	11,652	1.52
oltp_order_range	2,310	3,014	1.30
oltp_distinct_range	3,331	4,169	1.25
oltp_index_scan	3,303	5,072	1.54
select_random_points	10,007	14,119	1.41
select_random_ranges	3,929	5,028	1.28
covering_index_scan	7,437	9,840	1.32
groupby_scan	27,329	32,750	1.20
index_join	3,984	5,749	1.44
index_join_scan	1,860	2,980	1.60
types_table_scan	10,721	12,749	1.19
table_scan	115,738	131,126	1.13
oltp_read_only	93,991	121,576	1.29
Average			1.35

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	17,911	26,043	1.45
oltp_insert	13,643	19,607	1.44
oltp_update_index	33,011	50,777	1.54
oltp_update_non_index	23,448	35,327	1.51
oltp_delete_insert	31,504	41,748	1.33
oltp_write_only	13,185	20,950	1.59
types_delete_insert	16,258	20,498	1.26
oltp_read_write	52,440	82,867	1.58
Average			1.46

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	88,846	37,540	0.42
oltp_range_select	15,723	15,428	0.98
oltp_sum_range	14,795	14,277	0.96
oltp_order_range	3,171	4,467	1.41
oltp_distinct_range	4,092	5,612	1.37
oltp_index_scan	10,413	8,141	0.78
select_random_points	17,724	16,931	0.96
select_random_ranges	10,988	7,676	0.70
covering_index_scan	14,541	10,969	0.75
groupby_scan	28,214	34,534	1.22
index_join	7,738	8,008	1.03
index_join_scan	2,757	3,620	1.31
types_table_scan	11,419	12,849	1.13
table_scan	116,683	133,552	1.14
oltp_read_only	192,124	134,893	0.70
Average			0.99

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	20,979	29,349	1.40
oltp_insert	16,013	22,133	1.38
oltp_update_index	37,027	56,508	1.53
oltp_update_non_index	26,703	39,914	1.49
oltp_delete_insert	33,406	47,727	1.43
oltp_write_only	16,924	26,205	1.55
types_delete_insert	17,093	20,949	1.23
oltp_read_write	55,807	88,925	1.59
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	43,019	37,520	0.87
oltp_range_select	10,853	15,390	1.42
oltp_sum_range	10,028	14,301	1.43
oltp_order_range	2,598	4,465	1.72
oltp_distinct_range	3,613	5,618	1.55
oltp_index_scan	5,595	8,123	1.45
select_random_points	12,510	17,025	1.36
select_random_ranges	6,312	7,683	1.22
covering_index_scan	9,844	10,983	1.12
groupby_scan	27,557	34,605	1.26
index_join	5,221	8,022	1.54
index_join_scan	2,174	3,621	1.67
types_table_scan	10,908	12,784	1.17
table_scan	116,344	133,935	1.15
oltp_read_only	126,172	135,510	1.07
Average			1.33

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	73,060	78,096	1.07
oltp_insert_ac	74,761	102,925	1.38
oltp_update_index_ac	77,209	116,765	1.51
oltp_update_non_index_ac	70,860	94,185	1.33
oltp_delete_insert_ac	75,530	109,179	1.45
oltp_write_only_ac	77,093	106,938	1.39
types_delete_insert_ac	71,763	85,469	1.19
oltp_read_write_ac	81,240	109,757	1.35
Average			1.33

10000 rows, median of 21 invocations per test, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (1.8x individual, 1.5x average)

All tests within ceilings.

github-actions · 2026-05-20T22:34:37Z

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 1.8×; section averages gated at 1.5×.

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	23,353	29,139	1.25
oltp_range_select	9,844	13,661	1.39
oltp_sum_range	8,497	13,325	1.57
oltp_order_range	2,485	3,049	1.23
oltp_distinct_range	3,538	4,199	1.19
oltp_index_scan	3,196	4,257	1.33
select_random_points	11,328	14,354	1.27
select_random_ranges	4,097	4,985	1.22
covering_index_scan	7,289	9,205	1.26
groupby_scan	23,407	28,105	1.20
index_join	4,337	6,270	1.45
index_join_scan	1,658	2,653	1.60
types_table_scan	11,004	12,831	1.17
table_scan	8,376	10,589	1.26
oltp_read_only	100,797	128,130	1.27
Average			1.31

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	2,275	3,400	1.49
oltp_insert	16,208	25,157	1.55
oltp_update_index	38,722	50,600	1.31
oltp_update_non_index	28,703	33,484	1.17
oltp_delete_insert	28,532	42,612	1.49
oltp_write_only	13,066	19,119	1.46
types_delete_insert	15,655	18,512	1.18
oltp_read_write	54,737	79,299	1.45
Average			1.39

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	105,242	41,166	0.39
oltp_range_select	18,121	15,292	0.84
oltp_sum_range	16,853	14,690	0.87
oltp_order_range	3,350	3,512	1.05
oltp_distinct_range	4,455	4,708	1.06
oltp_index_scan	11,468	5,875	0.51
select_random_points	19,919	16,038	0.81
select_random_ranges	12,468	6,553	0.53
covering_index_scan	15,626	10,507	0.67
groupby_scan	24,520	28,631	1.17
index_join	8,606	7,292	0.85
index_join_scan	2,552	3,220	1.26
types_table_scan	11,934	12,978	1.09
table_scan	9,442	11,024	1.17
oltp_read_only	217,047	144,472	0.67
Average			0.86

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	4,101	4,301	1.05
oltp_insert	18,617	27,694	1.49
oltp_update_index	40,166	51,885	1.29
oltp_update_non_index	29,531	34,219	1.16
oltp_delete_insert	29,614	43,241	1.46
oltp_write_only	14,443	20,106	1.39
types_delete_insert	16,548	19,103	1.15
oltp_read_write	55,678	79,740	1.43
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	50,316	40,897	0.81
oltp_range_select	12,825	15,294	1.19
oltp_sum_range	11,312	14,727	1.30
oltp_order_range	2,791	3,584	1.28
oltp_distinct_range	3,839	4,710	1.23
oltp_index_scan	5,982	5,885	0.98
select_random_points	14,344	15,908	1.11
select_random_ranges	6,910	6,494	0.94
covering_index_scan	10,062	10,492	1.04
groupby_scan	23,768	28,416	1.20
index_join	5,653	7,258	1.28
index_join_scan	1,948	3,208	1.65
types_table_scan	11,269	12,970	1.15
table_scan	8,709	10,954	1.26
oltp_read_only	139,259	144,792	1.04
Average			1.16

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	54,431	65,178	1.20
oltp_insert_ac	55,043	79,736	1.45
oltp_update_index_ac	57,744	93,738	1.62
oltp_update_non_index_ac	52,184	73,202	1.40
oltp_delete_insert_ac	55,550	83,700	1.51
oltp_write_only_ac	54,273	84,225	1.55
types_delete_insert_ac	50,080	68,935	1.38
oltp_read_write_ac	60,566	91,057	1.50
Average			1.45

1000 rows, median of 13 invocations per test, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (1.8x individual, 1.5x average)

All tests within ceilings.

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 1.8×; section averages gated at 1.5×.

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	23,205	28,771	1.24
oltp_range_select	9,842	13,777	1.40
oltp_sum_range	8,440	13,401	1.59
oltp_order_range	2,461	3,076	1.25
oltp_distinct_range	3,530	4,233	1.20
oltp_index_scan	3,231	4,218	1.31
select_random_points	11,191	14,238	1.27
select_random_ranges	4,072	4,988	1.22
covering_index_scan	7,267	9,167	1.26
groupby_scan	23,252	28,162	1.21
index_join	4,370	6,229	1.43
index_join_scan	1,711	2,702	1.58
types_table_scan	11,016	12,798	1.16
table_scan	8,215	10,627	1.29
oltp_read_only	100,592	128,247	1.27
Average			1.31

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	2,323	3,380	1.46
oltp_insert	16,279	24,543	1.51
oltp_update_index	38,704	50,045	1.29
oltp_update_non_index	28,887	33,200	1.15
oltp_delete_insert	28,287	41,586	1.47
oltp_write_only	12,896	18,809	1.46
types_delete_insert	15,615	18,695	1.20
oltp_read_write	54,325	78,830	1.45
Average			1.37

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	104,339	40,770	0.39
oltp_range_select	18,183	15,264	0.84
oltp_sum_range	16,784	14,689	0.88
oltp_order_range	3,352	3,511	1.05
oltp_distinct_range	4,480	4,676	1.04
oltp_index_scan	11,488	5,830	0.51
select_random_points	19,629	15,794	0.80
select_random_ranges	12,407	6,460	0.52
covering_index_scan	15,624	10,528	0.67
groupby_scan	24,155	28,660	1.19
index_join	8,544	7,188	0.84
index_join_scan	2,576	3,186	1.24
types_table_scan	11,872	13,090	1.10
table_scan	9,238	10,880	1.18
oltp_read_only	207,808	143,911	0.69
Average			0.86

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	3,984	4,255	1.07
oltp_insert	18,280	26,935	1.47
oltp_update_index	39,580	50,189	1.27
oltp_update_non_index	29,654	34,398	1.16
oltp_delete_insert	28,864	42,038	1.46
oltp_write_only	13,843	19,611	1.42
types_delete_insert	16,326	19,020	1.17
oltp_read_write	54,983	79,451	1.45
Average			1.31

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	50,148	40,460	0.81
oltp_range_select	12,611	15,397	1.22
oltp_sum_range	11,235	14,716	1.31
oltp_order_range	2,762	3,495	1.27
oltp_distinct_range	3,840	4,666	1.22
oltp_index_scan	5,950	5,837	0.98
select_random_points	14,003	15,814	1.13
select_random_ranges	6,788	6,447	0.95
covering_index_scan	10,002	10,541	1.05
groupby_scan	23,565	28,299	1.20
index_join	5,741	7,230	1.26
index_join_scan	1,998	3,155	1.58
types_table_scan	11,404	12,997	1.14
table_scan	8,605	10,919	1.27
oltp_read_only	138,525	144,268	1.04
Average			1.16

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	53,569	63,977	1.19
oltp_insert_ac	54,465	80,352	1.48
oltp_update_index_ac	56,426	92,804	1.64
oltp_update_non_index_ac	52,361	69,786	1.33
oltp_delete_insert_ac	53,708	84,836	1.58
oltp_write_only_ac	53,863	84,013	1.56
types_delete_insert_ac	51,721	67,698	1.31
oltp_read_write_ac	58,528	92,385	1.58
Average			1.46

1000 rows, median of 13 invocations per test, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (1.8x individual, 1.5x average)

All tests within ceilings.

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 1.8×; section averages gated at 1.5×.

In-Memory

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	26,345	32,411	1.23
oltp_range_select	15,942	23,407	1.47
oltp_sum_range	14,772	21,782	1.47
oltp_order_range	3,105	4,112	1.32
oltp_distinct_range	4,161	5,293	1.27
oltp_index_scan	3,355	4,427	1.32
select_random_points	21,094	25,697	1.22
select_random_ranges	5,968	7,598	1.27
covering_index_scan	7,202	9,112	1.27
groupby_scan	28,130	35,363	1.26
index_join	5,738	7,968	1.39
index_join_scan	2,074	3,175	1.53
types_table_scan	10,991	12,913	1.17
table_scan	8,371	10,618	1.27
oltp_read_only	129,729	172,452	1.33
Average			1.32

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	2,376	3,301	1.39
oltp_insert	16,308	22,802	1.40
oltp_update_index	40,892	47,790	1.17
oltp_update_non_index	31,740	35,721	1.13
oltp_delete_insert	30,666	41,304	1.35
oltp_write_only	13,999	18,685	1.33
types_delete_insert	15,624	18,785	1.20
oltp_read_write	71,107	96,404	1.36
Average			1.29

File-Backed

Reads

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_point_select	101,965	43,703	0.43
oltp_range_select	24,405	25,123	1.03
oltp_sum_range	22,870	23,299	1.02
oltp_order_range	3,954	4,524	1.14
oltp_distinct_range	5,055	5,708	1.13
oltp_index_scan	11,698	5,884	0.50
select_random_points	29,619	27,302	0.92
select_random_ranges	14,318	8,911	0.62
covering_index_scan	15,699	10,396	0.66
groupby_scan	29,161	36,280	1.24
index_join	9,963	8,759	0.88
index_join_scan	2,933	3,661	1.25
types_table_scan	11,932	12,940	1.08
table_scan	9,452	11,028	1.17
oltp_read_only	246,799	188,256	0.76
Average			0.92

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert	4,172	4,198	1.01
oltp_insert	17,976	24,831	1.38
oltp_update_index	42,020	48,742	1.16
oltp_update_non_index	32,705	36,619	1.12
oltp_delete_insert	31,410	41,771	1.33
oltp_write_only	14,926	19,423	1.30
types_delete_insert	16,436	19,155	1.17
oltp_read_write	71,630	97,278	1.36
Average			1.23

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	53,010	44,057	0.83
oltp_range_select	18,652	25,564	1.37
oltp_sum_range	17,093	23,378	1.37
oltp_order_range	3,414	4,500	1.32
oltp_distinct_range	4,489	5,712	1.27
oltp_index_scan	6,106	5,898	0.97
select_random_points	23,681	27,243	1.15
select_random_ranges	8,564	8,881	1.04
covering_index_scan	10,039	10,374	1.03
groupby_scan	28,670	36,055	1.26
index_join	6,977	8,773	1.26
index_join_scan	2,418	3,684	1.52
types_table_scan	11,375	12,989	1.14
table_scan	8,779	10,994	1.25
oltp_read_only	167,960	187,376	1.12
Average			1.19

Writes

Test	SQLite (us)	Doltlite (us)	Multiplier
oltp_bulk_insert_ac	56,634	68,925	1.22
oltp_insert_ac	57,251	81,699	1.43
oltp_update_index_ac	57,871	90,921	1.57
oltp_update_non_index_ac	57,978	74,057	1.28
oltp_delete_insert_ac	61,631	86,084	1.40
oltp_write_only_ac	60,077	82,776	1.38
types_delete_insert_ac	56,338	71,749	1.27
oltp_read_write_ac	62,557	87,176	1.39
Average			1.37

1000 rows, median of 13 invocations per test, workload-only timing via host monotonic clock when available.

Performance Ceiling Check (1.8x individual, 1.5x average)

All tests within ceilings.

doltlite_revert_dirty.sh was written against the older "refuse only on touched-table dirty" rule. With the new staged-anywhere check, two tests need updating: * rv_dirty_same_refuses — expected error string updated from the old "cannot revert with uncommitted changes" to Dolt 2.0.5's "Your local changes would be overwritten by revert". * rv_staged_unrelated_* — was asserting that staged unrelated changes are folded into the revert commit. Under the new rule, any staged change refuses the revert (git's index-clean requirement), so flip the assertion: refuse, no new commit, target table unchanged, staged table still dirty. Case 1 (unstaged unrelated → allowed) and case 4 (clean → allowed) are unchanged; the new rule preserves those paths exactly. Verified locally: 14/14 pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

timsehn merged commit b57fb1e into master May 20, 2026
10 checks passed

timsehn deleted the fix/revert-match-dolt-2.0.5 branch May 21, 2026 01:42

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(revert): match Dolt 2.0.5 dirty-set rules (PR dolt#11073 + follow-up)#992

fix(revert): match Dolt 2.0.5 dirty-set rules (PR dolt#11073 + follow-up)#992
timsehn merged 2 commits into
masterfrom
fix/revert-match-dolt-2.0.5

timsehn commented May 20, 2026

Uh oh!

github-actions Bot commented May 20, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented May 20, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

timsehn commented May 20, 2026

Summary

What was wrong

What changed

References

Test plan

Uh oh!

github-actions Bot commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Sysbench-Style Benchmark: Doltlite vs SQLite

In-Memory

Reads

Writes

File-Backed

Reads

Writes

File-Backed (autocommit)

Reads

Writes

Performance Ceiling Check (1.8x individual, 1.5x average)

Uh oh!

github-actions Bot commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

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

In-Memory

Reads

Writes

File-Backed

Reads

Writes

File-Backed (autocommit)

Reads

Writes

Performance Ceiling Check (1.8x individual, 1.5x average)

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

In-Memory

Reads

Writes

File-Backed

Reads

Writes

File-Backed (autocommit)

Reads

Writes

Performance Ceiling Check (1.8x individual, 1.5x average)

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

In-Memory

Reads

Writes

File-Backed

Reads

Writes

File-Backed (autocommit)

Reads

Writes

Performance Ceiling Check (1.8x individual, 1.5x average)

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

github-actions Bot commented May 20, 2026 •

edited

Loading

github-actions Bot commented May 20, 2026 •

edited

Loading