test(perf): preserve deregistration trial pairing - #424
Merged
Eli Pinkerton (wallstop) merged 1 commit intoAug 16, 2026
Merged
Conversation
Contributor
Performance NumbersMeasured commit: Benchmark or harness paths changed, so historical deltas are non-comparable:
Historical Standalone deltaDelta direction: + is better; - is worse. Historical delta omitted because benchmark or harness code changed in this pull request. Current TargetMap evidenceThese raw rows are current-run diagnostics. Use fresh bracketed controls before accepting a micro-optimization. |
Eli Pinkerton (wallstop)
deleted the
perf/session-220-paired-deregistration-evidence
branch
August 16, 2026 01:45
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Science
The old diagnostic could combine four different host phases. The new marker retains all arms from one trial and reports
jointTrialSelection=true,sameTrialArms=true, the selected trial, and the fulltrialSequence.Four fresh-population invocations in one loaded Mono editor qualified only 2/4 samples. The final balanced eight-trial sample also rejected itself at 3.79% handler-excess spread. This PR therefore improves evidence quality but does not implement or claim acceptance for the exact-
MessageBusspecialization.Validation
npm run validate:allgit diff --checkThe only Unity console errors are pre-existing malformed
Assets/Plugins/Zenject/package.json.metahost-project errors outside this package.Refs #414.
Note
Low Risk
Changes are confined to perf benchmarks, contract tests, and documentation; no production dispatch or registration behavior is modified.
Overview
Replaces the deregistration H/B/B/H palindrome diagnostic that minimized four arms in separate seven-trial windows with
RunPairedDiagnostic, which runs eight trials, prepares all four fresh populations before timing, executes the arms back-to-back, and keeps the lowest complete palindrome so arms never come from different host phases. Preparation alternates forward/reverse; the structured log and tests recordjointTrialSelection,sameTrialArms,trialSequence, and fixed H/B/B/H execution order, while arithmetic-onlyAnalyzePalindromepaths no longer claim measured provenance.Adds
DeregistrationAttributionPalindromeSample, injects benchmark registry isolation viaCreateState(dispose on failed construction), and extends contract coverage for floor selection and arm indexing. Perf methodology and campaign decisions document the accepted joint method, diagnostic peak-memory tradeoff, and that this still does not authorize the exact-MessageBuscandidate.Reviewed by Cursor Bugbot for commit 00a360c. Bugbot is set up for automated code reviews on this repo. Configure here.