perf(ast): ASCII fast path in identifierNeedsQuoting, hash-free isReservedName [W5-04] by ajitpratap0 · Pull Request #945 · ajitpratap0/GoSQLX

ajitpratap0 · 2026-11-03T17:19:02Z

Summary

identifierNeedsQuoting: 256-entry byte-class table for ASCII names; the first non-ASCII byte falls back to the unchanged rune loop (identifierNeedsQuotingRunes).
isReservedName: first-letter/length index of [16]byte entries derived at init from reservedNameWords (which stays the single word list), array-compare scan, zero allocations. This deviates from the PLAN label "switch-based": a generated index keeps one source of truth at the same speed.
No API or behaviour change; AST.SQL() and formatter output byte-identical.

Test Plan

New pkg/sql/ast/quote_fastpath_test.go: frozen reference copies, exhaustive 0-2 byte and piece-combination equivalence, reserved-name case/affix/exhaustive 3-byte equivalence, index invariants, FuzzQuoteFastPath (60s, 944k execs, clean), zero-alloc pins, micro benchmarks.
Mutation proof (each reverted): M1-M3 TestIdentifierNeedsQuoting_MatchesReference; M4 TestIdentifierByteClass_NonASCII; M5, M7 TestIsReservedName_MatchesReference; M6 TestReservedNameIndex_Invariants + TestIsReservedName_MatchesReference.
Byte identity: SQL dump of all 5,422 corpus renderings identical to origin/main (git-archive base). task goldens clean, no golden changes.
task check (after merging origin/main), task test:cbinding, go test ./pkg/compatibility/..., staticcheck on pkg/sql/ast/..., targeted -race run: all pass.
Local A/B (task bench:ab, Apple M4, shared machine, indicative only; spread up to +/-223%): BenchmarkRender geomean -9.78%; sql/select_where -17%, sql/join -15%, sql/cte -21%, readable/delete -9.7%; sql/complex and sql/insert no change.
CI perf gate on 8996df9 (comment 6612262939; layout-randomized, 20 layouts/arm, runner CPU Intel(R) Xeon(R) 6975P-C): PASS. No time regression; allocs/op and B/op unchanged on every gated benchmark; instructions/op between -7.34% (e2e/cte) and +0.06% (e2e/simple_select), e2e/complex -4.32%; A/A' control clean (0 of 52 benchmarks moved more than 3%, suite geomean -0.08%, p 0.369). Low power list (9): ParseError/long_prefix/validate, ParseE2E/delete, LintRule/L010, ParseE2E/simple_select, ParseRW/clickhouse/medium, ParseRW/snowflake/medium, Tokenize/scratch/insert, Tokenize/scratch/complex, LintRW/oracle; none is in this change's blast radius, and allocs/op, B/op and instructions/op guard them.
The one significant gated improvement is the guard PLAN section 5 names, BenchmarkNormalize (renders through the same helpers): simple_select -2.85%, join -2.35% (both p<0.001, status ok), complex -4.94% (p<0.001, flagged 'faster'); allocs/op and B/op unchanged. Under PLAN 5.3 rule 5 this is the amd64 evidence of a real gain, so the PR is not closed.
Refreshed PR gate after merging origin/main (base 6e86a6b, head abde69d, AMD EPYC 9V74): PASS, all checks green. Normalize again significant and faster: simple_select -1.82%, join -2.27%, complex -3.32% (p<0.001, A/A' -0.85%..-0.47%), allocs/op and B/op unchanged.
Target benchmark pkg/formatter BenchmarkRender (not in gate-suite.txt), via the layout-randomized matrix route (perf-gate.yml -f mode=matrix -f rounds=20 -f layouts=20, run 41706707217, AMD EPYC 9V74, GOGC=off, pinned CPU). The job hit the 230-minute limit before the A/A' stage and the report step, so its report is ERROR; I ran scripts/perf/gate on the 16 completed layouts per arm of the raw gcoff files instead (no A/A' control for this run). The base of that run (6d29d68) carried the L007 keyword_case change that this branch lacked, so ignore its Lint rows; Render rows are unaffected. Render/sql: simple_select -10.96%, select_where -17.56%, join -14.77%, aggregate -14.35%, subquery -15.16%, cte -19.89%, window -18.57%, complex -17.31%, insert -10.70%, update -13.08%, delete -12.78% (all p<0.001, allocs 1 -> 1, B/op +0.00%). Render/readable: simple_select -7.65%, select_where -10.98%, join -8.23%, aggregate -9.80%, subquery -8.63%, cte -9.17%, window -8.14%, complex -9.01%, insert -7.95%, update -5.72%, delete -7.57% (all p<0.001, allocs/op and B/op unchanged). Targets met: render/sql <= -7% (worst row -10.70%) and render/readable <= -3% (worst row -5.72%).
PLAN deviation: a generated first-letter/length index instead of a switch-based isReservedName (one word list stays the source of truth).

Breaking Changes

None.

🤖 Generated with Claude Code

https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

…ervedName [W5-04] Byte-class table for ASCII names (rune loop only for non-ASCII), and a first-letter/length index derived from reservedNameWords replaces the map lookup. SQL() output is byte-identical. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

vercel · 2026-11-03T17:19:05Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
gosqlx-website	Ready	Preview	Nov 4, 2026 12:03am UTC

github-actions · 2026-11-03T17:51:45Z

Claude finished @ajitpratap0's task in 22s —— View job

Review of PR #945 (ASCII fast path for identifier quoting)

I read the diff and traced the logic by hand. I did not run the tests or benchmarks. The sandbox blocked the go test command, so the check below is by reading only.

Verdict: Looks correct. I found no blocking issues.

identifierNeedsQuoting (pkg/sql/ast/sql.go)

The byte-class state machine matches the rune loop for ASCII input.
- Digit-led parts: a word byte (including *) in a digit-led part returns true, as the rune loop does. Digits keep the part bare, which is what ClickHouse d.1 needs.
- $: it is rejected at the start of a part or in a digit-led part, and allowed later, as in the rune loop.
- Everything else: other ASCII bytes return true.
Falling back to identifierNeedsQuotingRunes(name) on the first byte ≥ 0x80 is safe. The fallback restarts from the beginning, so no state carries over. Invalid UTF-8 also goes down the rune path, which is the conservative choice.
The empty-name check is duplicated in both functions. That is harmless.

isReservedName (pkg/sql/ast/quote_names.go)

Deriving the index at init from reservedNameWords keeps one source of truth.
The init-time validation panics on bad words, so a future typo in the list fails loudly instead of silently missing the index.
uint16 and uint8 are comfortably wide enough for the entry count and bucket size.
The [16]byte array compare requires up to be zero-padded beyond len(name). up is a fresh zero value, so that holds.

Minor nits (non-blocking)

reservedNameEntry and reservedNameIndex have no doc comments, unlike the code around them. A short comment on the bucket layout ([firstLetter][length] → start, n) would help. The anonymous struct type is repeated twice in the initializer, so a named reservedBucket type would be tidier.
The CHANGELOG entry says "measured numbers await the CI verdict". Fill in the real numbers once the perf gate reports, as the repo's "measure, don't assume" rule expects.
In the changelog line, "Identifier quoting [W5-04]" puts the ticket tag in a different spot than the neighbouring entries, which put it at the end. This is cosmetic.

Tests

Test coverage is strong:
- frozen reference copies,
- exhaustive short-input equivalence,
- a fuzz test,
- zero-allocation pins,
- mutation proofs.
The FuzzQuoteFastPath run (944k execs) and the byte-identical corpus dump give good confidence that behaviour is unchanged.

Performance and security

No allocations and no map hashing remain on the hot path.
There is no new injection surface. Names that need quoting still go through writeQuotedIdentifier, and the fast path only returns "bare" for the same inputs as before.

…ut [W5-04] Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

github-actions · 2026-11-03T18:31:58Z

Perf gate: PASS (shadow mode: report only, this check never fails)

Base 6e86a6b8b9b1, head abde69dc9aa9. 20 interleaved rounds, 20 randomized code layouts per arm (-randlayout seeds 417437552412..417437552431; A/A' 417437552432..417437552451), benchtime 100ms, -cpu 1, GOMEMLIMIT 64MiB, pinned to CPU 3 (SMT siblings 2-3) with taskset.

Runner CPU: AMD EPYC 9V74 80-Core Processor (x86_64). Base and head ran interleaved in this job, so the tables show their ratios only; absolute timings (raw files in the artifact) hold for this CPU and this job alone and are never compared across jobs.

2 benchmark(s) faster by 3% or more

Rules: each arm runs 20 randomized code layouts (-ldflags=-randlayout); time regresses at >= +3% when significant at a false discovery rate of 0.01 over the suite (Benjamini-Hochberg on exact Mann-Whitney U p-values of the per-layout medians, so a regression must hold across layouts, not within one binary), or a suite geomean >= +1.5% (per-layout suite geomeans, same test); allocs/op must not increase; B/op must not rise more than 2%; instructions/op must not rise more than 1.5%. The A/A' control runs the base revision under a disjoint set of layout seeds; when more than 10% of benchmarks move over 3% in it, the run is noisy: re-run once, then inconclusive (neutral).

Gated suite (GOGC=off, GOMAXPROCS=1)

benchmark	time	p	A/A'	spread	B/op	allocs/op	status
`ParseE2E/simple_select`	-0.12%	0.850	-1.13%	1.78%	+0.00%	5 → 5	ok
`ParseE2E/select_where`	+0.26%	0.490	+0.09%	1.04%	+0.00%	18 → 18	ok
`ParseE2E/join`	-0.14%	0.808	+0.12%	1.59%	+0.00%	17 → 17	ok
`ParseE2E/aggregate`	+0.12%	0.914	-0.28%	3.97%	+0.00%	18 → 18	ok
`ParseE2E/subquery`	+0.90%	0.086	-0.06%	1.18%	+0.00%	14 → 14	ok
`ParseE2E/cte`	+0.24%	0.525	+0.37%	1.21%	+0.00%	21 → 21	ok
`ParseE2E/window`	-0.34%	0.330	-0.31%	0.97%	+0.00%	11 → 11	ok
`ParseE2E/complex`	+0.28%	0.561	-0.23%	1.55%	+0.00%	49 → 49	ok
`ParseE2E/insert`	-0.05%	0.968	-0.66%	1.63%	+0.00%	17 → 17	ok
`ParseE2E/update`	-0.05%	0.882	-0.43%	1.73%	+0.00%	13 → 13	ok
`ParseE2E/delete`	-0.26%	0.705	-0.64%	1.57%	+0.00%	5 → 5	ok
`ParseRW/postgresql/medium`	+0.13%	0.925	-1.33%	2.53%	+0.00%	646 → 646	ok
`ParseRW/mysql/medium`	+1.36%	0.091	-0.11%	2.58%	+0.00%	368 → 368	ok
`ParseRW/mariadb/medium`	+0.38%	0.512	-0.39%	1.28%	+0.00%	528 → 528	ok
`ParseRW/sqlserver/medium`	+0.63%	0.102	+0.24%	0.44%	+0.00%	927 → 927	ok
`ParseRW/oracle/medium`	+0.56%	0.072	-0.13%	0.72%	+0.00%	1090 → 1090	ok
`ParseRW/sqlite/medium`	+0.48%	0.341	-0.25%	1.22%	+0.00%	347 → 347	ok
`ParseRW/snowflake/medium`	+0.95%	0.076	+0.29%	1.61%	+0.00%	776 → 776	ok
`ParseRW/clickhouse/medium`	+0.59%	0.327	+0.06%	1.17%	+0.00%	383 → 383	ok
`ParseFresh/simple_select`	-0.08%	0.818	+0.16%	1.15%	+0.00%	6 → 6	ok
`ParseFresh/join`	+0.22%	0.507	+0.49%	1.17%	+0.00%	18 → 18	ok
`ParseFresh/complex`	+0.32%	0.394	+0.20%	1.42%	+0.00%	50 → 50	ok
`PoolGetPut/serial`	-3.53%	<0.001	+0.17%	0.30%	+0.00%	0 → 0	faster
`Tokenize/scratch/join`	-0.04%	0.989	-0.71%	1.38%	+0.00%	0 → 0	ok
`Tokenize/scratch/complex`	+0.24%	0.766	-0.48%	1.47%	+0.00%	0 → 0	ok
`Tokenize/scratch/insert`	+0.53%	0.665	+0.11%	3.15%	+0.00%	0 → 0	ok
`Normalize/simple_select`	-1.82%	<0.001	-0.85%	1.32%	+0.00%	7 → 7	ok
`Normalize/join`	-2.27%	<0.001	-0.82%	0.76%	+0.00%	23 → 23	ok
`Normalize/complex`	-3.32%	<0.001	-0.47%	0.75%	+0.00%	68 → 68	faster
`TransformParse/simple_select`	+0.12%	0.892	+0.00%	0.86%	+0.00%	5 → 5	ok
`TransformParse/join`	-0.08%	0.946	+0.38%	1.05%	+0.00%	17 → 17	ok
`TransformParse/complex`	+0.48%	0.561	+0.10%	2.31%	+0.00%	49 → 49	ok
`LintParallel/rules`	+0.86%	0.414	+0.94%	3.03%	+0.00%	68929.5 → 68929	ok
`LintParallel/lint`	+0.36%	0.698	-0.53%	1.16%	-0.00%	147508 → 147508	ok
`LintRW/postgresql`	+0.36%	0.429	+0.01%	1.05%	+0.00%	16338 → 16338	ok
`LintRW/mysql`	-0.03%	0.968	+0.51%	1.72%	-0.00%	20352.5 → 20352	ok
`LintRW/mariadb`	-0.66%	0.289	+0.23%	1.67%	+0.00%	18484 → 18484	ok
`LintRW/sqlserver`	+0.57%	0.253	+0.80%	1.59%	+0.00%	15940 → 15940	ok
`LintRW/oracle`	-0.02%	0.989	+0.47%	2.62%	+0.00%	19674 → 19674	ok
`LintRW/sqlite`	-0.57%	0.277	-0.71%	1.41%	+0.00%	18302 → 18302	ok
`LintRW/snowflake`	+0.11%	0.779	-0.09%	1.40%	+0.00%	17765 → 17765	ok
`LintRW/clickhouse`	+0.72%	0.265	+0.94%	2.30%	+0.00%	20645 → 20645	ok
`LintRule/L007`	-0.32%	0.698	-0.29%	1.96%	+0.00%	1194 → 1194	ok
`LintRule/L010`	-0.01%	0.989	+0.15%	1.54%	+0.00%	0 → 0	ok
`ParseError/syntax_mid/parse`	-0.32%	0.441	-0.24%	1.59%	+0.00%	12 → 12	ok
`ParseError/syntax_mid/validate`	-0.18%	0.871	-0.33%	1.77%	+0.00%	12 → 12	ok
`ParseError/syntax_end/parse`	+0.18%	0.787	-0.56%	2.06%	+0.00%	16 → 16	ok
`ParseError/syntax_end/validate`	-0.28%	0.655	-0.93%	2.16%	+0.00%	16 → 16	ok
`ParseError/tokenize/parse`	-1.36%	0.204	-0.60%	2.76%	+0.00%	4 → 4	ok
`ParseError/tokenize/validate`	-1.45%	0.126	+0.02%	1.95%	+0.00%	21 → 21	ok
`ParseError/long_prefix/parse`	-0.05%	0.904	-0.29%	1.22%	+0.00%	219 → 219	ok
`ParseError/long_prefix/validate`	+0.44%	0.114	+0.27%	0.68%	+0.00%	219 → 219	ok
geomean	-0.10%	0.398	-0.13%

time and A/A' are Hodges-Lehmann shifts over the per-layout medians. spread: the base arm's robust standard deviation across its layouts (1.4826 x MAD, relative), i.e. how much this benchmark moves between builds that differ only in function placement, plus run-to-run noise; a change much smaller than it cannot be resolved in one run.

A/A' control (base revision, disjoint layout seeds): 0 of 52 benchmarks moved more than 3%; 0 would have been flagged by the time rule; suite geomean -0.13%, p 0.758.

Low power

1 benchmark(s) have under 80% estimated power to be flagged on their own for a +5% slowdown (their spread across layouts is too wide for 20 layouts at the Benjamini-Hochberg level alpha/m = 0.000192). A slowdown below the last column usually goes unflagged there; instructions/op, allocs/op and the suite geomean still guard them.

benchmark	spread	power at +5%	slowdown caught 80% of the time
`ParseE2E/aggregate`	3.97%	53%	+6.0%

Power is estimated with the normal approximation for the U test (Noether): power = Phi(ln(1+shift)/s * sqrt(K/2) * sqrt(3/pi) - z(1 - alpha/(2m))), with s the spread and K the layouts per arm.

Instructions per operation (cmd/perfdriver)

case	instructions/op	status
`e2e/aggregate`	+0.04%	ok
`e2e/complex`	+0.00%	ok
`e2e/cte`	-0.02%	ok
`e2e/delete`	-0.04%	ok
`e2e/insert`	-0.08%	ok
`e2e/join`	-0.04%	ok
`e2e/select_where`	-0.01%	ok
`e2e/simple_select`	+0.02%	ok
`e2e/subquery`	+0.06%	ok
`e2e/update`	-0.01%	ok
`e2e/window`	+0.00%	ok
`rw/postgresql/medium`	+0.06%	ok

Secondary metric: default GC, GOMAXPROCS=1 (informational)

benchmark	time	p
`ParseE2E/simple_select`	-0.52%	0.427
`ParseE2E/select_where`	+0.08%	0.970
`ParseE2E/join`	-0.01%	1.000
`ParseE2E/aggregate`	-1.30%	0.199
`ParseE2E/subquery`	+0.09%	0.971
`ParseE2E/cte`	-0.24%	0.684
`ParseE2E/window`	-0.47%	0.579
`ParseE2E/complex`	-0.19%	0.796
`ParseE2E/insert`	-0.07%	0.910
`ParseE2E/update`	+0.68%	0.353
`ParseE2E/delete`	+0.48%	0.596
`ParseRW/postgresql/medium`	+0.71%	0.579
`ParseRW/mysql/medium`	+1.80%	0.123
`ParseRW/mariadb/medium`	+1.21%	0.190
`ParseRW/sqlserver/medium`	+0.79%	0.481
`ParseRW/oracle/medium`	+1.95%	0.023
`ParseRW/sqlite/medium`	+1.33%	0.123
`ParseRW/snowflake/medium`	+1.08%	0.280
`ParseRW/clickhouse/medium`	+0.33%	0.739
`ParseFresh/simple_select`	-0.70%	0.105
`ParseFresh/join`	+1.06%	0.165
`ParseFresh/complex`	+0.48%	0.631
`PoolGetPut/serial`	-3.57%	<0.001
`Tokenize/scratch/join`	+0.04%	0.970
`Tokenize/scratch/complex`	-0.01%	1.000
`Tokenize/scratch/insert`	-0.97%	0.436
`Normalize/simple_select`	-1.07%	0.218
`Normalize/join`	-2.26%	0.063
`Normalize/complex`	-1.95%	0.019
`TransformParse/simple_select`	-0.18%	0.791
`TransformParse/join`	+0.02%	1.000
`TransformParse/complex`	+0.04%	0.971
`LintParallel/rules`	+0.20%	0.971
`LintParallel/lint`	-0.20%	0.971
`LintRW/postgresql`	+0.70%	0.739
`LintRW/mysql`	-0.06%	0.971
`LintRW/mariadb`	+0.65%	0.247
`LintRW/sqlserver`	+0.28%	0.853
`LintRW/oracle`	+0.86%	0.579
`LintRW/sqlite`	-0.11%	0.912
`LintRW/snowflake`	+1.06%	0.280
`LintRW/clickhouse`	-0.62%	0.481
`LintRule/L007`	-1.08%	0.315
`LintRule/L010`	+0.73%	0.481
`ParseError/syntax_mid/parse`	+0.50%	0.450
`ParseError/syntax_mid/validate`	+0.39%	0.256
`ParseError/syntax_end/parse`	+1.60%	0.041
`ParseError/syntax_end/validate`	+1.14%	0.112
`ParseError/tokenize/parse`	-0.73%	0.739
`ParseError/tokenize/validate`	+1.73%	0.212
`ParseError/long_prefix/parse`	+1.29%	0.315
`ParseError/long_prefix/validate`	+0.38%	0.631
geomean	+0.14%

Environment

date:        2026-11-04T00:06:06Z
base:        6e86a6b8b9b1e7dbb91268edd0972fbe357e773b
head:        working tree of abde69dc9aa994949468d18ec83579d4eed1870f
cpu model:   AMD EPYC 9V74 80-Core Processor (x86_64)
go:          go version go1.26.9 linux/amd64
kernel:      Linux 7.0.0-1017-azure x86_64
layouts:     20 per arm, -ldflags=-randlayout seeds 417437552412..417437552431 (base, head), 417437552432..417437552451 (A/A')
pinning:     pinned to CPU 3 (SMT siblings 2-3) with taskset
gcoff env:   GOGC=off GOMEMLIMIT=64MiB
clocksource: tsc
load:        7.33, 2.55, 0.92
instr count: cachegrind
perf probe:  Error: No supported events found. The instructions:u event is not supported. 

Architecture:                            x86_64
CPU(s):                                  4
Model name:                              AMD EPYC 9V74 80-Core Processor
Thread(s) per core:                      2
Core(s) per socket:                      2
Socket(s):                               1
Hypervisor vendor:                       Microsoft
L1d cache:                               64 KiB (2 instances)
L1i cache:                               64 KiB (2 instances)
L2 cache:                                2 MiB (2 instances)
L3 cache:                                32 MiB (1 instance)

benchstat (primary metric, samples pooled across layouts: informational, the verdict uses the per-layout test)

goos: linux
goarch: amd64
pkg: github.com/ajitpratap0/GoSQLX/pkg/fingerprint
cpu: AMD EPYC 9V74 80-Core Processor                
                        │    base     │                head                │
                        │   sec/op    │   sec/op     vs base               │
Normalize/simple_select   2.366µ ± 1%   2.323µ ± 1%  -1.80% (p=0.000 n=20)
Normalize/join            6.924µ ± 0%   6.767µ ± 1%  -2.27% (p=0.000 n=20)
Normalize/complex         18.21µ ± 1%   17.58µ ± 1%  -3.41% (p=0.000 n=20)
geomean                   6.681µ        6.514µ       -2.50%

                        │     base     │                 head                  │
                        │     B/op     │     B/op      vs base                 │
Normalize/simple_select     896.0 ± 0%     896.0 ± 0%       ~ (p=1.000 n=20) ¹
Normalize/join            2.211Ki ± 0%   2.211Ki ± 0%       ~ (p=1.000 n=20) ¹
Normalize/complex         6.117Ki ± 0%   6.117Ki ± 0%       ~ (p=1.000 n=20) ¹
geomean                   2.279Ki        2.279Ki       +0.00%
¹ all samples are equal

                        │    base    │                head                 │
                        │ allocs/op  │ allocs/op   vs base                 │
Normalize/simple_select   7.000 ± 0%   7.000 ± 0%       ~ (p=1.000 n=20) ¹
Normalize/join            23.00 ± 0%   23.00 ± 0%       ~ (p=1.000 n=20) ¹
Normalize/complex         68.00 ± 0%   68.00 ± 0%       ~ (p=1.000 n=20) ¹
geomean                   22.20        22.20       +0.00%
¹ all samples are equal

pkg: github.com/ajitpratap0/GoSQLX/pkg/gosqlx
                                │    base     │                head                │
                                │   sec/op    │   sec/op     vs base               │
ParseE2E/simple_select            1.710µ ± 1%   1.714µ ± 1%       ~ (p=0.846 n=20)
ParseE2E/select_where             3.498µ ± 1%   3.510µ ± 0%       ~ (p=0.490 n=20)
ParseE2E/join                     4.609µ ± 1%   4.599µ ± 1%       ~ (p=0.805 n=20)
ParseE2E/aggregate                3.530µ ± 4%   3.539µ ± 2%       ~ (p=0.909 n=20)
ParseE2E/subquery                 3.150µ ± 1%   3.179µ ± 1%       ~ (p=0.085 n=20)
ParseE2E/cte                      4.412µ ± 1%   4.442µ ± 1%       ~ (p=0.525 n=20)
ParseE2E/window                   2.906µ ± 1%   2.895µ ± 1%       ~ (p=0.330 n=20)
ParseE2E/complex                  12.00µ ± 2%   11.99µ ± 1%       ~ (p=0.560 n=20)
ParseE2E/insert                   2.136µ ± 1%   2.134µ ± 1%       ~ (p=0.963 n=20)
ParseE2E/update                   1.889µ ± 1%   1.889µ ± 1%       ~ (p=0.878 n=20)
ParseE2E/delete                   1.180µ ± 2%   1.176µ ± 1%       ~ (p=0.703 n=20)
ParseRW/postgresql/medium         153.6µ ± 3%   154.3µ ± 4%       ~ (p=0.925 n=20)
ParseRW/mysql/medium              86.52µ ± 2%   88.07µ ± 3%       ~ (p=0.091 n=20)
ParseRW/mariadb/medium            119.3µ ± 1%   119.9µ ± 1%       ~ (p=0.512 n=20)
ParseRW/sqlserver/medium          222.1µ ± 0%   223.7µ ± 1%       ~ (p=0.102 n=20)
ParseRW/oracle/medium             271.4µ ± 0%   273.1µ ± 1%       ~ (p=0.072 n=20)
ParseRW/sqlite/medium             68.51µ ± 1%   68.90µ ± 1%       ~ (p=0.341 n=20)
ParseRW/snowflake/medium          174.9µ ± 1%   176.3µ ± 1%       ~ (p=0.076 n=20)
ParseRW/clickhouse/medium         82.75µ ± 1%   82.80µ ± 2%       ~ (p=0.327 n=20)
ParseFresh/simple_select          7.077µ ± 1%   7.084µ ± 1%       ~ (p=0.815 n=20)
ParseFresh/join                   11.26µ ± 1%   11.31µ ± 1%       ~ (p=0.507 n=20)
ParseFresh/complex                20.02µ ± 1%   20.05µ ± 1%       ~ (p=0.394 n=20)
PoolGetPut/serial                 76.01n ± 0%   73.35n ± 0%  -3.50% (p=0.000 n=20)
ParseError/syntax_mid/parse       2.469µ ± 1%   2.464µ ± 1%       ~ (p=0.441 n=20)
ParseError/syntax_mid/validate    2.178µ ± 2%   2.182µ ± 1%       ~ (p=0.867 n=20)
ParseError/syntax_end/parse       2.840µ ± 2%   2.849µ ± 1%       ~ (p=0.784 n=20)
ParseError/syntax_end/validate    2.639µ ± 3%   2.631µ ± 3%       ~ (p=0.654 n=20)
ParseError/tokenize/parse         873.0n ± 2%   858.7n ± 2%       ~ (p=0.204 n=20)
ParseError/tokenize/validate      2.773µ ± 1%   2.728µ ± 2%       ~ (p=0.126 n=20)
ParseError/long_prefix/parse      61.37µ ± 1%   61.35µ ± 1%       ~ (p=0.904 n=20)
ParseError/long_prefix/validate   57.90µ ± 1%   58.11µ ± 1%       ~ (p=0.114 n=20)
geomean                           9.251µ        9.253µ       +0.03%

                                │     base     │                head                 │
                                │     B/s      │     B/s       vs base               │
ParseE2E/simple_select            14.50Mi ± 1%   14.47Mi ± 1%       ~ (p=0.836 n=20)
ParseE2E/select_where             28.09Mi ± 1%   27.99Mi ± 0%       ~ (p=0.490 n=20)
ParseE2E/join                     26.07Mi ± 1%   26.13Mi ± 1%       ~ (p=0.825 n=20)
ParseE2E/aggregate                22.70Mi ± 4%   22.64Mi ± 2%       ~ (p=0.920 n=20)
ParseE2E/subquery                 24.82Mi ± 1%   24.60Mi ± 1%       ~ (p=0.087 n=20)
ParseE2E/cte                      26.80Mi ± 1%   26.62Mi ± 1%       ~ (p=0.516 n=20)
ParseE2E/window                   30.85Mi ± 1%   30.97Mi ± 1%       ~ (p=0.311 n=20)
ParseE2E/complex                  30.05Mi ± 2%   30.06Mi ± 1%       ~ (p=0.560 n=20)
ParseE2E/insert                   33.94Mi ± 1%   33.96Mi ± 1%       ~ (p=0.952 n=20)
ParseE2E/update                   27.76Mi ± 1%   27.78Mi ± 0%       ~ (p=0.836 n=20)
ParseE2E/delete                   42.04Mi ± 2%   42.16Mi ± 1%       ~ (p=0.723 n=20)
ParseRW/postgresql/medium         23.37Mi ± 4%   23.25Mi ± 4%       ~ (p=0.920 n=20)
ParseRW/mysql/medium              31.28Mi ± 2%   30.73Mi ± 3%       ~ (p=0.090 n=20)
ParseRW/mariadb/medium            29.72Mi ± 1%   29.57Mi ± 1%       ~ (p=0.516 n=20)
ParseRW/sqlserver/medium          29.95Mi ± 0%   29.75Mi ± 1%       ~ (p=0.109 n=20)
ParseRW/oracle/medium             30.88Mi ± 0%   30.69Mi ± 1%       ~ (p=0.073 n=20)
ParseRW/sqlite/medium             30.15Mi ± 1%   29.98Mi ± 1%       ~ (p=0.351 n=20)
ParseRW/snowflake/medium          29.07Mi ± 1%   28.84Mi ± 1%       ~ (p=0.080 n=20)
ParseRW/clickhouse/medium         27.23Mi ± 1%   27.22Mi ± 2%       ~ (p=0.317 n=20)
ParseError/syntax_mid/parse       18.54Mi ± 1%   18.58Mi ± 1%       ~ (p=0.425 n=20)
ParseError/syntax_mid/validate    21.01Mi ± 1%   20.99Mi ± 1%       ~ (p=0.846 n=20)
ParseError/syntax_end/parse       13.10Mi ± 2%   13.06Mi ± 1%       ~ (p=0.774 n=20)
ParseError/syntax_end/validate    14.09Mi ± 3%   14.14Mi ± 3%       ~ (p=0.654 n=20)
ParseError/tokenize/parse         50.25Mi ± 2%   51.09Mi ± 2%       ~ (p=0.204 n=20)
ParseError/tokenize/validate      15.82Mi ± 1%   16.08Mi ± 2%       ~ (p=0.119 n=20)
ParseError/long_prefix/parse      37.64Mi ± 1%   37.65Mi ± 1%       ~ (p=0.909 n=20)
ParseError/long_prefix/validate   39.89Mi ± 1%   39.75Mi ± 1%       ~ (p=0.116 n=20)
geomean                           26.40Mi        26.37Mi       -0.14%

                                │      base      │                 head                  │
                                │      B/op      │     B/op      vs base                 │
ParseE2E/simple_select              816.0 ± 0%       816.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/select_where             1.523Ki ± 0%     1.523Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/join                     1.836Ki ± 0%     1.836Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/aggregate                1.812Ki ± 0%     1.812Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/subquery                 1.828Ki ± 0%     1.828Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/cte                      2.336Ki ± 0%     2.336Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/window                   1.281Ki ± 0%     1.281Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/complex                  5.025Ki ± 0%     5.025Ki ± 0%       ~ (p=0.695 n=20)
ParseE2E/insert                     824.0 ± 0%       824.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/update                     736.0 ± 0%       736.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/delete                     400.0 ± 0%       400.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/postgresql/medium         65.90Ki ± 0%     65.90Ki ± 0%       ~ (p=0.638 n=20)
ParseRW/mysql/medium              42.37Ki ± 0%     42.37Ki ± 0%       ~ (p=0.768 n=20)
ParseRW/mariadb/medium            60.60Ki ± 0%     60.60Ki ± 0%       ~ (p=0.091 n=20)
ParseRW/sqlserver/medium          83.67Ki ± 0%     83.67Ki ± 0%       ~ (p=0.274 n=20)
ParseRW/oracle/medium             104.8Ki ± 0%     104.8Ki ± 0%       ~ (p=0.877 n=20)
ParseRW/sqlite/medium             38.12Ki ± 0%     38.12Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/snowflake/medium          77.63Ki ± 0%     77.63Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/clickhouse/medium         34.82Ki ± 0%     34.82Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseFresh/simple_select            896.0 ± 0%       896.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseFresh/join                   1.914Ki ± 0%     1.914Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseFresh/complex                5.102Ki ± 0%     5.102Ki ± 0%       ~ (p=1.000 n=20) ¹
PoolGetPut/serial                   0.000 ± 0%       0.000 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_mid/parse         680.0 ± 0%       680.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_mid/validate      680.0 ± 0%       680.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_end/parse       1.430Ki ± 0%     1.430Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_end/validate    1.430Ki ± 0%     1.430Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseError/tokenize/parse           272.0 ± 0%       272.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/tokenize/validate      1.758Ki ± 0%     1.758Ki ± 0%       ~ (p=1.000 n=20) ¹
ParseError/long_prefix/parse      22.75Ki ± 0%     22.75Ki ± 0%       ~ (p=0.161 n=20)
ParseError/long_prefix/validate   22.68Ki ± 0%     22.68Ki ± 0%       ~ (p=1.000 n=20) ¹
geomean                                        ²                 +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                                │     base      │                 head                 │
                                │   allocs/op   │  allocs/op   vs base                 │
ParseE2E/simple_select             5.000 ± 0%      5.000 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/select_where              18.00 ± 0%      18.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/join                      17.00 ± 0%      17.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/aggregate                 18.00 ± 0%      18.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/subquery                  14.00 ± 0%      14.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/cte                       21.00 ± 0%      21.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/window                    11.00 ± 0%      11.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/complex                   49.00 ± 0%      49.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/insert                    17.00 ± 0%      17.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/update                    13.00 ± 0%      13.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseE2E/delete                    5.000 ± 0%      5.000 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/postgresql/medium          646.0 ± 0%      646.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/mysql/medium               368.0 ± 0%      368.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/mariadb/medium             528.0 ± 0%      528.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/sqlserver/medium           927.0 ± 0%      927.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/oracle/medium             1.090k ± 0%     1.090k ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/sqlite/medium              347.0 ± 0%      347.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/snowflake/medium           776.0 ± 0%      776.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseRW/clickhouse/medium          383.0 ± 0%      383.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseFresh/simple_select           6.000 ± 0%      6.000 ± 0%       ~ (p=1.000 n=20) ¹
ParseFresh/join                    18.00 ± 0%      18.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseFresh/complex                 50.00 ± 0%      50.00 ± 0%       ~ (p=1.000 n=20) ¹
PoolGetPut/serial                  0.000 ± 0%      0.000 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_mid/parse        12.00 ± 0%      12.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_mid/validate     12.00 ± 0%      12.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_end/parse        16.00 ± 0%      16.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/syntax_end/validate     16.00 ± 0%      16.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/tokenize/parse          4.000 ± 0%      4.000 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/tokenize/validate       21.00 ± 0%      21.00 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/long_prefix/parse       219.0 ± 0%      219.0 ± 0%       ~ (p=1.000 n=20) ¹
ParseError/long_prefix/validate    219.0 ± 0%      219.0 ± 0%       ~ (p=1.000 n=20) ¹
geomean                                       ²                +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                          │    base     │                head                │
                          │  sec/stmt   │  sec/stmt    vs base               │
ParseRW/postgresql/medium   12.80µ ± 3%   12.86µ ± 4%       ~ (p=0.925 n=20)
ParseRW/mysql/medium        10.82µ ± 2%   11.01µ ± 3%       ~ (p=0.091 n=20)
ParseRW/mariadb/medium      13.26µ ± 1%   13.32µ ± 1%       ~ (p=0.512 n=20)
ParseRW/sqlserver/medium    10.58µ ± 0%   10.65µ ± 1%       ~ (p=0.101 n=20)
ParseRW/oracle/medium       11.31µ ± 0%   11.38µ ± 1%       ~ (p=0.072 n=20)
ParseRW/sqlite/medium       11.42µ ± 1%   11.48µ ± 1%       ~ (p=0.337 n=20)
ParseRW/snowflake/medium    13.45µ ± 1%   13.56µ ± 1%       ~ (p=0.075 n=20)
ParseRW/clickhouse/medium   11.82µ ± 1%   11.83µ ± 2%       ~ (p=0.324 n=20)
geomean                     11.89µ        11.97µ       +0.69%

pkg: github.com/ajitpratap0/GoSQLX/pkg/linter
                   │     base     │                head                 │
                   │    sec/op    │    sec/op     vs base               │
LintParallel/rules   20.51m ±  2%   20.79m ±  2%       ~ (p=0.414 n=20)
LintParallel/lint    40.08m ±  1%   40.37m ±  2%       ~ (p=0.698 n=20)
LintRW/postgresql    5.393m ±  1%   5.423m ±  1%       ~ (p=0.429 n=20)
LintRW/mysql         5.606m ±  1%   5.612m ±  1%       ~ (p=0.968 n=20)
LintRW/mariadb       5.046m ±  1%   5.025m ±  2%       ~ (p=0.289 n=20)
LintRW/sqlserver     4.778m ±  1%   4.809m ±  1%       ~ (p=0.253 n=20)
LintRW/oracle        5.667m ±  2%   5.658m ±  1%       ~ (p=0.989 n=20)
LintRW/sqlite        4.694m ±  1%   4.657m ±  1%       ~ (p=0.277 n=20)
LintRW/snowflake     4.819m ±  1%   4.818m ±  1%       ~ (p=0.779 n=20)
LintRW/clickhouse    5.380m ±  2%   5.426m ±  1%       ~ (p=0.265 n=20)
LintRule/L007        2.826m ±  1%   2.817m ±  2%       ~ (p=0.698 n=20)
LintRule/L010        974.0µ ± 16%   972.6µ ± 16%       ~ (p=0.989 n=20)
geomean              5.684m         5.695m        +0.20%

                  │     base     │                head                 │
                  │     B/s      │     B/s       vs base               │
LintRW/postgresql   5.255Mi ± 1%   5.221Mi ± 1%       ~ (p=0.409 n=20)
LintRW/mysql        5.836Mi ± 1%   5.832Mi ± 1%       ~ (p=1.000 n=20)
LintRW/mariadb      6.046Mi ± 1%   6.070Mi ± 2%       ~ (p=0.267 n=20)
LintRW/sqlserver    6.485Mi ± 1%   6.442Mi ± 1%       ~ (p=0.285 n=20)
LintRW/oracle       7.010Mi ± 2%   7.019Mi ± 1%       ~ (p=0.995 n=20)
LintRW/sqlite       6.113Mi ± 1%   6.161Mi ± 1%       ~ (p=0.256 n=20)
LintRW/snowflake    6.475Mi ± 1%   6.475Mi ± 1%       ~ (p=0.783 n=20)
LintRW/clickhouse   6.032Mi ± 2%   5.980Mi ± 1%       ~ (p=0.261 n=20)
geomean             6.137Mi        6.130Mi       -0.12%

                   │      base      │                 head                  │
                   │      B/op      │     B/op      vs base                 │
LintParallel/rules   5.559Mi ± 0%     5.559Mi ± 0%       ~ (p=0.888 n=20)
LintParallel/lint    25.53Mi ± 0%     25.53Mi ± 0%       ~ (p=0.984 n=20)
LintRW/postgresql    2.801Mi ± 0%     2.801Mi ± 0%       ~ (p=0.963 n=20)
LintRW/mysql         3.537Mi ± 0%     3.537Mi ± 0%       ~ (p=0.147 n=20)
LintRW/mariadb       3.252Mi ± 0%     3.252Mi ± 0%       ~ (p=0.394 n=20)
LintRW/sqlserver     2.838Mi ± 0%     2.838Mi ± 0%       ~ (p=0.100 n=20)
LintRW/oracle        3.477Mi ± 0%     3.477Mi ± 0%       ~ (p=0.560 n=20)
LintRW/sqlite        3.066Mi ± 0%     3.066Mi ± 0%       ~ (p=0.417 n=20)
LintRW/snowflake     3.083Mi ± 0%     3.083Mi ± 0%       ~ (p=0.176 n=20)
LintRW/clickhouse    3.474Mi ± 0%     3.474Mi ± 0%       ~ (p=0.878 n=20)
LintRule/L007        92.98Ki ± 0%     92.98Ki ± 0%       ~ (p=1.000 n=20) ¹
LintRule/L010          0.000 ± 0%       0.000 ± 0%       ~ (p=1.000 n=20) ¹
geomean                           ²                 +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                   │     base      │                 head                 │
                   │   allocs/op   │  allocs/op   vs base                 │
LintParallel/rules   68.93k ± 0%     68.93k ± 0%       ~ (p=0.585 n=20)
LintParallel/lint    147.5k ± 0%     147.5k ± 0%       ~ (p=0.559 n=20)
LintRW/postgresql    16.34k ± 0%     16.34k ± 0%       ~ (p=0.342 n=20)
LintRW/mysql         20.35k ± 0%     20.35k ± 0%       ~ (p=0.751 n=20)
LintRW/mariadb       18.48k ± 0%     18.48k ± 0%       ~ (p=1.000 n=20)
LintRW/sqlserver     15.94k ± 0%     15.94k ± 0%       ~ (p=0.176 n=20)
LintRW/oracle        19.67k ± 0%     19.67k ± 0%       ~ (p=1.000 n=20)
LintRW/sqlite        18.30k ± 0%     18.30k ± 0%       ~ (p=0.320 n=20)
LintRW/snowflake     17.77k ± 0%     17.77k ± 0%       ~ (p=1.000 n=20) ¹
LintRW/clickhouse    20.64k ± 0%     20.64k ± 0%       ~ (p=1.000 n=20)
LintRule/L007        1.194k ± 0%     1.194k ± 0%       ~ (p=1.000 n=20) ¹
LintRule/L010         0.000 ± 0%      0.000 ± 0%       ~ (p=1.000 n=20) ¹
geomean                          ²                -0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                  │     base     │                head                 │
                  │   sec/stmt   │   sec/stmt    vs base               │
LintRW/postgresql   17.68µ ±  1%   17.78µ ±  1%       ~ (p=0.425 n=20)
LintRW/mysql        13.54µ ±  1%   13.56µ ±  1%       ~ (p=0.962 n=20)
LintRW/mariadb      13.28µ ±  1%   13.22µ ±  2%       ~ (p=0.286 n=20)
LintRW/sqlserver    17.25µ ±  1%   17.36µ ±  1%       ~ (p=0.253 n=20)
LintRW/oracle       15.40µ ±  2%   15.38µ ±  1%       ~ (p=0.989 n=20)
LintRW/sqlite       13.49µ ±  1%   13.38µ ±  1%       ~ (p=0.274 n=20)
LintRW/snowflake    14.74µ ±  1%   14.73µ ±  1%       ~ (p=0.784 n=20)
LintRW/clickhouse   13.65µ ±  2%   13.77µ ±  1%       ~ (p=0.265 n=20)
LintRule/L007       1.005µ ±  1%   1.001µ ±  2%       ~ (p=0.683 n=20)
LintRule/L010       346.2n ± 16%   345.8n ± 16%       ~ (p=0.963 n=20)
geomean             7.766µ         7.767µ        +0.02%

pkg: github.com/ajitpratap0/GoSQLX/pkg/sql/tokenizer
                         │    base     │                head                │
                         │   sec/op    │   sec/op     vs base               │
Tokenize/scratch/join      1.397µ ± 1%   1.392µ ± 1%       ~ (p=0.984 n=20)
Tokenize/scratch/complex   3.675µ ± 1%   3.691µ ± 1%       ~ (p=0.763 n=20)
Tokenize/scratch/insert    694.6n ± 2%   701.8n ± 2%       ~ (p=0.664 n=20)
geomean                    1.528µ        1.533µ       +0.37%

                         │     base     │                head                 │
                         │     B/s      │     B/s       vs base               │
Tokenize/scratch/join      86.03Mi ± 1%   86.36Mi ± 1%       ~ (p=1.000 n=20)
Tokenize/scratch/complex   98.11Mi ± 1%   97.68Mi ± 1%       ~ (p=0.753 n=20)
Tokenize/scratch/insert    104.3Mi ± 2%   103.3Mi ± 2%       ~ (p=0.678 n=20)
geomean                    95.85Mi        95.51Mi       -0.36%

                         │     base     │                head                 │
                         │     B/op     │    B/op     vs base                 │
Tokenize/scratch/join      0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=20) ¹
Tokenize/scratch/complex   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=20) ¹
Tokenize/scratch/insert    0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=20) ¹
geomean                               ²               +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                         │     base     │                head                 │
                         │  allocs/op   │ allocs/op   vs base                 │
Tokenize/scratch/join      0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=20) ¹
Tokenize/scratch/complex   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=20) ¹
Tokenize/scratch/insert    0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=20) ¹
geomean                               ²               +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

pkg: github.com/ajitpratap0/GoSQLX/pkg/transform
                             │    base     │                head                │
                             │   sec/op    │   sec/op     vs base               │
TransformParse/simple_select   1.637µ ± 1%   1.637µ ± 1%       ~ (p=0.888 n=20)
TransformParse/join            5.083µ ± 1%   5.089µ ± 0%       ~ (p=0.941 n=20)
TransformParse/complex         13.26µ ± 1%   13.30µ ± 1%       ~ (p=0.560 n=20)
geomean                        4.797µ        4.802µ       +0.11%

                             │     base     │                 head                  │
                             │     B/op     │     B/op      vs base                 │
TransformParse/simple_select     816.0 ± 0%     816.0 ± 0%       ~ (p=1.000 n=20) ¹
TransformParse/join            1.836Ki ± 0%   1.836Ki ± 0%       ~ (p=1.000 n=20) ¹
TransformParse/complex         5.023Ki ± 0%   5.023Ki ± 0%       ~ (p=1.000 n=20) ¹
geomean                        1.944Ki        1.944Ki       +0.00%
¹ all samples are equal

                             │    base    │                head                 │
                             │ allocs/op  │ allocs/op   vs base                 │
TransformParse/simple_select   5.000 ± 0%   5.000 ± 0%       ~ (p=1.000 n=20) ¹
TransformParse/join            17.00 ± 0%   17.00 ± 0%       ~ (p=1.000 n=20) ¹
TransformParse/complex         49.00 ± 0%   49.00 ± 0%       ~ (p=1.000 n=20) ¹
geomean                        16.09        16.09       +0.00%
¹ all samples are equal

Run, raw data and benchstat

github-actions · 2026-11-03T18:47:56Z

Claude finished @ajitpratap0's task in 18s —— View job

Review of PR #945 (ASCII fast path for identifier quoting)

I read the diff against origin/main. I did not run tests or benchmarks. This review is from reading the code only.

Verdict: Looks correct. I found no blocking issues.

identifierNeedsQuoting (pkg/sql/ast/sql.go)

The byte-class state machine matches the rune loop for ASCII input.
- Word bytes, including *, in a digit-led part return true.
- Digits keep a part bare, which keeps ClickHouse d.1 working.
- $ is rejected at the start of a part or inside a digit-led part, and allowed elsewhere.
- Any other ASCII byte returns true.
The first byte ≥ 0x80 hands the whole name to identifierNeedsQuotingRunes. That restarts from the beginning, so no state carries over. Invalid UTF-8 also takes this path, which is the conservative choice.

isReservedName (pkg/sql/ast/quote_names.go)

Building the index at init from reservedNameWords keeps a single word list.
A bad word panics at init, so a typo in the list fails loudly.
Bucket indexing is safe. len(name) is capped at 16, so [maxReservedNameLen+1] is large enough.
The [16]byte compare works because up starts zeroed, which pads the unused bytes.
uint16 and uint8 are wide enough for the current list size.
The earlier nits about doc comments and a named bucket type are already addressed (reservedNameBucket).

Remaining nits (non-blocking)

CHANGELOG.md still says "measured numbers await the CI verdict". Replace it with the real perf-gate numbers once they are in, per the repo's "measure, don't assume" rule.
The ticket tag [W5-04] is missing from that entry. Neighbouring entries put it at the end.

Tests

Coverage is strong:
- frozen reference copies,
- exhaustive short-input equivalence,
- a fuzz test,
- zero-allocation pins,
- mutation proofs.
The byte-identical corpus dump (5,422 renderings) gives good confidence that behaviour is unchanged.

Performance and security

The hot path has no allocations and no map hashing.
There is no new injection surface. Names that need quoting still go through the same quoting path, and the fast path returns "bare" only for the same inputs as before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

github-actions · 2026-11-03T19:38:56Z

Claude finished @ajitpratap0's task in 15s —— View job

Review of PR #945 (ASCII fast path for identifier quoting)

I read the diff against origin/main. I did not run tests or benchmarks, so this is a review of the code only.

Verdict: Looks correct. I found no blocking issues.

identifierNeedsQuoting (pkg/sql/ast/sql.go)

The byte-class state machine matches the documented rules:
- A word byte (letter, _ or *) in a digit-led part returns true.
- Digits keep a part bare, which keeps ClickHouse d.1 working.
- . resets the part state.
- $ is rejected at the start of a part or inside a digit-led part, and allowed elsewhere.
- Any other ASCII byte returns true.
The first byte ≥ 0x80 hands the whole name to identifierNeedsQuotingRunes. That restarts from the beginning, so no state carries over. Invalid UTF-8 also goes down the rune path.
The empty-name check appears in both functions. That is harmless.

isReservedName (pkg/sql/ast/quote_names.go)

The index is built at init from reservedNameWords, so the word list stays the single source of truth.
A bad word panics at init, so a typo in the list fails loudly.
Bucket indexing is safe: len(name) is capped at maxReservedNameLen, and the array has maxReservedNameLen+1 slots.
up is zero-initialised, so the [16]byte compare holds for names shorter than 16 bytes.
uint16 and uint8 are wide enough for the current list. The init-time check does not guard against overflow if the list ever grows past 255 words per bucket. That is not a realistic risk.
The earlier nit about doc comments and a named bucket type is addressed (reservedNameBucket).

Nits (non-blocking)

The reservedNameEntries, reservedNameIndex declaration is a long single-line signature. Naming the index type (for example type reservedNameIndexTable [26][maxReservedNameLen + 1]reservedNameBucket) would read better.
The [W5-04] tag sits at the end of the CHANGELOG entry, which matches the neighbouring entries. The "await CI verdict" wording is gone and real numbers are in.
The CHANGELOG entry is honest that BenchmarkRender is outside the gated suite. Consider adding it to the perf-gate suite in a follow-up, so the PLAN targets can be confirmed on amd64. The CI gate here only shows no regression: Normalize/* is about 2–3% faster, and allocs/op and B/op are unchanged.

Tests

Coverage is strong:
- frozen reference copies,
- exhaustive short-input equivalence,
- a fuzz test,
- zero-allocation pins,
- mutation proofs,
- a byte-identical corpus dump (5,422 renderings).

Performance and security

The hot path has no allocations and no map hashing.
The fast path returns "bare" only for the same inputs as before, so there is no new injection surface.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

github-actions · 2026-11-04T00:02:27Z

Claude finished @ajitpratap0's task in 14s —— View job

Review of PR #945 (ASCII fast path for identifier quoting)

I read the diff against origin/main (sql.go, quote_names.go, CHANGELOG.md). I did not run tests or benchmarks.

Verdict: Looks correct. I found no blocking issues.

identifierNeedsQuoting (pkg/sql/ast/sql.go)

The byte-class state machine matches the documented rules:
- A word byte (letter, _ or *) in a digit-led part returns true.
- Digits keep a part bare, so ClickHouse d.1 still works.
- . resets the part state.
- $ is rejected at the start of a part or in a digit-led part, and allowed elsewhere.
- Any other ASCII byte returns true.
The first byte ≥ 0x80 hands the whole name to identifierNeedsQuotingRunes. It restarts from the beginning, so no state carries over. Invalid UTF-8 takes the same path.

isReservedName (pkg/sql/ast/quote_names.go)

The index is built at init from reservedNameWords, so the word list stays the single source of truth.
A bad word panics at init, so a typo in the list fails loudly.
Indexing is safe. len(name) is capped at maxReservedNameLen, and first is checked to be a–z before it is used as an index.
up is zero-initialised, so the [16]byte compare holds for names shorter than 16 bytes.

Nits (non-blocking)

The uint16 bucket start and uint8 count are not checked at init. Overflow is not realistic with the current list. A one-line panic guard would make a future list growth fail loudly.
The reservedNameEntries, reservedNameIndex declaration has a long one-line signature. A named type for the index table would read better.
The empty-name check appears in both identifierNeedsQuoting and the rune fallback. This is harmless.

CHANGELOG

The entry gives amd64 numbers for BenchmarkRender, so the PLAN targets are backed by data. It also states that run 41706707217 had no A/A' control. The PR body still says the gated suite has no BenchmarkRender, so consider updating it to match.
The PR gate shows no regression. Normalize/* is 2–5% faster, and allocs/op and B/op are unchanged.

Tests

Coverage is strong:
- frozen reference copies,
- exhaustive short-input equivalence,
- a fuzz test,
- zero-allocation pins,
- mutation proofs,
- a byte-identical corpus dump.

Performance and security

The hot path has no allocations and no map hashing.
The fast path returns "bare" only for the same inputs as before, so there is no new injection surface.

Test and others added 2 commits November 3, 2026 22:46

merge origin/main

0bb6bff

github-actions Bot added documentation Improvements or additions to documentation ast AST node changes labels Nov 3, 2026

refactor(ast): name reserved-name index bucket type and document layo…

2badfa6

…ut [W5-04] Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

vercel Bot deployed to Preview November 3, 2026 18:16

View deployment

docs(changelog): record W5-04 measurements

8996df9

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

vercel Bot deployed to Preview November 3, 2026 19:35

View deployment

merge origin/main

8c8a9f8

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DnbRnTFErhoMJdAjVEeCg2

vercel Bot deployed to Preview November 4, 2026 00:03

View deployment

Navigation Menu

Repository navigation

perf(ast): ASCII fast path in identifierNeedsQuoting, hash-free isReservedName [W5-04] - #945

perf(ast): ASCII fast path in identifierNeedsQuoting, hash-free isReservedName [W5-04]#945
ajitpratap0 wants to merge 5 commits into
mainfrom
perf/w5-04-quote-ascii-fastpath

ajitpratap0 commented Nov 3, 2026 •

edited

Loading

Uh oh!

vercel Bot commented Nov 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Nov 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Nov 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Nov 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Nov 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Nov 4, 2026 •

edited

Loading

Uh oh!

This branch was successfully deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

ajitpratap0 commented Nov 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test Plan

Breaking Changes

Uh oh!

vercel Bot commented Nov 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Nov 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review of PR #945 (ASCII fast path for identifier quoting)

Uh oh!

github-actions Bot commented Nov 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Perf gate: PASS (shadow mode: report only, this check never fails)

Gated suite (GOGC=off, GOMAXPROCS=1)

Low power

Instructions per operation (cmd/perfdriver)

Uh oh!

github-actions Bot commented Nov 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review of PR #945 (ASCII fast path for identifier quoting)

Uh oh!

github-actions Bot commented Nov 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review of PR #945 (ASCII fast path for identifier quoting)

Uh oh!

github-actions Bot commented Nov 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review of PR #945 (ASCII fast path for identifier quoting)

Uh oh!

This branch was successfully deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

ajitpratap0 commented Nov 3, 2026 •

edited

Loading

vercel Bot commented Nov 3, 2026 •

edited

Loading

github-actions Bot commented Nov 3, 2026 •

edited

Loading

github-actions Bot commented Nov 3, 2026 •

edited

Loading

github-actions Bot commented Nov 3, 2026 •

edited

Loading

github-actions Bot commented Nov 3, 2026 •

edited

Loading

github-actions Bot commented Nov 4, 2026 •

edited

Loading