Return SubSequence from SQLCommenter.getFirstWord to avoid per-inject substring by dougqh · Pull Request #11736 · DataDog/dd-trace-java

dougqh · 2026-06-24T22:15:19Z

What Does This Do

Updates SQLCommenter.getFirstWord to return a SubSequence instead of a sub-String.
By using SubSequence, we avoid the sub-string allocation and often the SubSequence will be optimized away completely via escape analysis.

Motivation

Reduce work in application threads
Reduce JDBC integration allocation

Additional Notes

SQLCommenter.getFirstWord returns a zero-copy SubSequence instead of sql.substring(...). Its result is only consumed by startsWith("{") / equalsIgnoreCase("call") in inject, so the per-inject substring allocation was pure waste. The returned view carries a "transient — do not retain" warning (it pins its backing String).

It also carries SubSequence.startsWith/equalsIgnoreCase (retracted from #10640 so they land with their motivating consumer rather than as speculative API).

SQLCommenterGetFirstWordBenchmark (JDK 17, @Threads(8), @Fork(5), -prof gc)

	throughput	`gc.alloc.rate.norm`
before (`substring`)	258.1M ± 21.0M ops/s	48 B/op
after (`SubSequence`)	508.0M ± 21.6M ops/s	~0 (10⁻⁴)

Escape analysis fully elides the view in the transient consumption: 48 B/op → ~0, ~2× throughput. (The win is contingent on non-escape — confirming why the view must not be retained.)

🤖 Generated with Claude Code

dd-octo-sts · 2026-06-24T22:38:39Z

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	13.96 s	13.92 s	[-0.4%; +0.9%] (no difference)
startup:insecure-bank:tracing:Agent	12.84 s	13.00 s	[-1.9%; -0.5%] (maybe better)
startup:petclinic:appsec:Agent	17.44 s	17.12 s	[+0.9%; +2.7%] (maybe worse)
startup:petclinic:iast:Agent	17.41 s	17.41 s	[-1.1%; +1.2%] (no difference)
startup:petclinic:profiling:Agent	17.37 s	17.32 s	[-0.8%; +1.4%] (no difference)
startup:petclinic:sca:Agent	16.82 s	17.42 s	[-7.7%; +1.0%] (no difference)
startup:petclinic:tracing:Agent	16.52 s	16.62 s	[-1.7%; +0.5%] (no difference)

Commit: e66f238a · CI Pipeline · Benchmarking Platform UI

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.

dougqh · 2026-06-25T11:35:28Z

Benchmark results (zulu-17, JDK 17.0.7, `@Threads(8)`, `@Fork(5)`, `-prof gc`)

SQLCommenterGetFirstWordBenchmark.firstWordCheck:

	throughput	gc.alloc.rate.norm
before (substring)	258.1M ± 21.0M ops/s	48 B/op
after (SubSequence)	508.0M ± 21.6M ops/s	~0 B/op (10⁻⁴)

The SubSequence view is escape-analysis–elided in the transient firstWord check, so the old substring's 48 B/op (a String + its backing array) drops to ~0 and throughput rises ~2× at @Threads(8). @Fork(5) tightens the earlier bimodal @Fork(2) spread. Benchmark Javadoc refreshed in 89ca7b8.

…comparisons) These are the ops a real classification parse needs but CharSequence lacks — surfaced by porting SQLCommenter.getFirstWord (firstWord.startsWith("{"), firstWord.equalsIgnoreCase("call")). Lets transient parse views be compared without materializing a String. + JUnit 5 tests. Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

… substring getFirstWord's result is only consumed by startsWith("{") and equalsIgnoreCase("call"), so materializing a substring on every inject() was pure waste. Return a zero-copy SubSequence view instead and use its startsWith/equalsIgnoreCase. Behavior is unchanged (91 SQLCommenterTest cases pass); the test compares getFirstWord(sql).toString(). This is the motivating consumer for SubSequence.startsWith/equalsIgnoreCase (retracted from #10640 and carried here so they land with their use case). Co-Authored-By: Claude Opus 4.8 <[email protected]>

@threads

Benchmarks the getFirstWord first-word scan, consuming the result via the real transient pattern (startsWith, returning the boolean) so EA isn't faked away. @threads(8), @fork(2), -prof gc: the old substring allocated 48 B/op per call; the SubSequence view is fully EA-elided (~0 B/op), ~1.6x throughput. Numbers in the class Javadoc. Co-Authored-By: Claude Opus 4.8 <[email protected]>

zulu-17 @fork(5), -prof gc: 258.1M -> 508.0M ops/s (~2x), 48 -> ~0 B/op. @fork(5) tightens the earlier bimodal @fork(2) spread. Co-Authored-By: Claude Opus 4.8 <[email protected]>

ygree · 2026-06-29T23:15:06Z

+   * for large SQL). Consume it in place and discard; if the value must be retained, call {@link
+   * SubSequence#toString()} to detach a standalone copy.
+   */
+  protected static SubSequence getFirstWord(String sql) {


Looks good for the generic SubSequence case to have extra helper methods such as startsWith and equalsIgnoreCase. Looking back at the concrete getFirstWord use cases, though, it could be much simpler: I would eliminate getFirstWord completely, since it does not seem to add much value here.

It is only used to skip whitespace and check whether the next token is either { or a case-insensitive call followed by whitespace. That could be checked directly in a small amount of code.

Fair point. I've been erring on the side of keeping the changes 1-to-1 and very mechanical, but I do agree that further simplification might make sense here.

dd-octo-sts · 2026-06-30T18:25:44Z

/merge

gh-worker-devflow-routing-ef8351 · 2026-06-30T18:25:48Z

View all feedbacks in Devflow UI.

2026-06-30 18:25:48 UTC ℹ️ Start processing command /merge

2026-06-30 18:25:52 UTC ℹ️ MergeQueue: pull request added to the queue

The expected merge time in master is approximately 2h (p90).

2026-06-30 19:24:13 UTC ℹ️ MergeQueue: This merge request was merged

The #11736 CharSequence (charAt-loop) versions are superseded by the String-delegating overloads here; String-literal callers (SQLCommenter) bind to the String overloads. Removes the redundant pair + their tests. Co-Authored-By: Claude Opus 4.8 <[email protected]>

Add window-bounded String overloads to SubSequence equals/equalsIgnoreCase/startsWith/endsWith/indexOf take a String and delegate to String's region/offset methods (regionMatches, startsWith, indexOf) instead of a per-char CharSequence loop. Each guards against this view's [beginIndex, endIndex) window first so the delegated read stays in range, then reuses the JDK's backing-array compare (Latin1 fast path / intrinsics). equals(Object) now routes Strings through the fast path, keeping the charAt loop only for non-String CharSequences. Co-Authored-By: Claude Opus 4.8 <[email protected]> Add char overloads to SubSequence: startsWith/endsWith/indexOf Single-character leading/trailing/search checks (e.g. a leading '{' or a trailing ';') read charAt(beginIndex)/charAt(endIndex-1) or delegate to String.indexOf(int, from), each bounded to the [beginIndex, endIndex) window. indexOf(char) returns a window-relative offset or -1. Co-Authored-By: Claude Opus 4.8 <[email protected]> Add lastIndexOf overloads; split equals into equals(String)/contentEquals - lastIndexOf(String) and lastIndexOf(char), window-bounded like indexOf, returning a window-relative offset. - Restructure equality to mirror String's API: equals(String) is the region-compare fast path, contentEquals(CharSequence) is the general char-by-char comparison, and equals(Object) dispatches String -> the fast path, any other CharSequence -> contentEquals. This keeps two equal-content views equal() while giving String args the fast path. Co-Authored-By: Claude Opus 4.8 <[email protected]> Compute SubSequence.hashCode allocation-free over the window Replace toString().hashCode() with the String hash polynomial evaluated directly over [beginIndex, endIndex). Same value (so equals/hashCode stay consistent), but hashing a view no longer materializes a substring -- preserving the zero-copy property the class exists for. Co-Authored-By: Claude Opus 4.8 <[email protected]> Drop redundant CharSequence equalsIgnoreCase/startsWith from SubSequence The #11736 CharSequence (charAt-loop) versions are superseded by the String-delegating overloads here; String-literal callers (SQLCommenter) bind to the String overloads. Removes the redundant pair + their tests. Co-Authored-By: Claude Opus 4.8 <[email protected]> Fix SubSequence.subSequence(start,end) end-index overshoot The CharSequence contract treats start/end as offsets in this view's coordinates, so absolute end is beginIndex+end, not beginIndex+start+end (which overshoots by start; only correct when start==0). Latent since test including the nested case the bug broke worst. Co-Authored-By: Claude Opus 4.8 <[email protected]> Add contentEquals(String) fast path; suppress intentional equals asymmetry Add a String-specialized contentEquals(String) using String.regionMatches, avoiding the CharSequence path's virtual charAt dispatch, and make equals(String) a thin alias for it so contentEquals is the primary content-comparison API. Suppress the SpotBugs EQ_CHECK_FOR_OPERAND_NOT_COMPATIBLE_WITH_THIS on equals(Object) -- the cross-type view equality is intentional and documented. Co-Authored-By: Claude Opus 4.8 <[email protected]> Merge branch 'master' into dougqh/subsequence-string-methods Co-authored-by: devflow.devflow-routing-intake <[email protected]>

dougqh added tag: performance Performance related changes tag: no release notes Changes to exclude from release notes type: refactoring inst: jdbc JDBC instrumentation tag: ai generated Largely based on code generated by an AI or LLM labels Jun 24, 2026

dougqh and others added 4 commits June 29, 2026 11:06

Refresh SQLCommenterGetFirstWordBenchmark results to JDK 17 @fork(5)

2f648c7

zulu-17 @fork(5), -prof gc: 258.1M -> 508.0M ops/s (~2x), 48 -> ~0 B/op. @fork(5) tightens the earlier bimodal @fork(2) spread. Co-Authored-By: Claude Opus 4.8 <[email protected]>

dougqh force-pushed the dougqh/sqlcommenter-getfirstword-subseq branch from 89ca7b8 to 2f648c7 Compare June 29, 2026 15:10

dougqh marked this pull request as ready for review June 29, 2026 17:04

dougqh requested review from a team as code owners June 29, 2026 17:04

dougqh requested review from ygree and removed request for a team June 29, 2026 17:04

ygree reviewed Jun 29, 2026

View reviewed changes

ygree approved these changes Jun 29, 2026

View reviewed changes

Merge branch 'master' into dougqh/sqlcommenter-getfirstword-subseq

e66f238

dougqh enabled auto-merge June 30, 2026 17:32

This comment has been minimized.

Sign in to view

dougqh added this pull request to the merge queue Jun 30, 2026

github-merge-queue Bot removed this pull request from the merge queue due to failed status checks Jun 30, 2026

gh-worker-dd-mergequeue-cf854d Bot merged commit 015a24f into master Jun 30, 2026
583 checks passed

gh-worker-dd-mergequeue-cf854d Bot deleted the dougqh/sqlcommenter-getfirstword-subseq branch June 30, 2026 19:24

github-actions Bot added this to the 1.64.0 milestone Jun 30, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Return SubSequence from SQLCommenter.getFirstWord to avoid per-inject substring#11736

Return SubSequence from SQLCommenter.getFirstWord to avoid per-inject substring#11736
gh-worker-dd-mergequeue-cf854d[bot] merged 5 commits into
masterfrom
dougqh/sqlcommenter-getfirstword-subseq

dougqh commented Jun 24, 2026 •

edited

Loading

Uh oh!

dd-octo-sts Bot commented Jun 24, 2026 •

edited

Loading

Uh oh!

dougqh commented Jun 25, 2026

Uh oh!

ygree Jun 29, 2026

Uh oh!

dougqh Jun 30, 2026

Uh oh!

This comment has been minimized.

dd-octo-sts Bot commented Jun 30, 2026

Uh oh!

gh-worker-devflow-routing-ef8351 Bot commented Jun 30, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

dougqh commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What Does This Do

Motivation

Additional Notes

Uh oh!

dd-octo-sts Bot commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🟢 Java Benchmark SLOs — All performance SLOs passed

Uh oh!

dougqh commented Jun 25, 2026

Benchmark results (zulu-17, JDK 17.0.7, @Threads(8), @Fork(5), -prof gc)

Uh oh!

ygree Jun 29, 2026

Choose a reason for hiding this comment

Uh oh!

dougqh Jun 30, 2026

Choose a reason for hiding this comment

Uh oh!

This comment has been minimized.

dd-octo-sts Bot commented Jun 30, 2026

Uh oh!

gh-worker-devflow-routing-ef8351 Bot commented Jun 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dougqh commented Jun 24, 2026 •

edited

Loading

dd-octo-sts Bot commented Jun 24, 2026 •

edited

Loading

Benchmark results (zulu-17, JDK 17.0.7, `@Threads(8)`, `@Fork(5)`, `-prof gc`)

gh-worker-devflow-routing-ef8351 Bot commented Jun 30, 2026 •

edited

Loading