test(engine): drop non-portable 5k EPS floor in DurationBound (IB-21) - #90
Merged
Merged
Conversation
The TestRun_DurationBound/5k_eps sustained-EPS lower floor (generated ≥ 0.85× eps·after) is not portable to constrained CI: a shared 2-core runner cannot physically push 5k EPS through generate→pipeline→send, so it lands far below the floor (ratio ~0.34) and failed the v0.4.1 release Test gate, though it passes on a fast dev box (~1.04). Remove the 5k lower-floor assertion outright (founder Option (c)); retain it only at the 1k subtest (10 req/s, trivially sustainable on 2 cores). No capability probe / measured-send-rate / skip machinery. The IB-19 stop.after overrun guard (wall ≤ after+1s AND generated ≤ 1.15× eps·after) is preserved universally at BOTH 1k and 5k — it still fails-before on be2d195 (1k 3.63×/9.25s, 5k 1.32×/6.02s) and passes-after on main. A governance comment marks those assertions as the IB-19 regression guard that must not be hidden behind testing.Short(), so ci.yml and release.yml keep catching the class pre-merge. Test-only: no internal/engine product code changed; IB-19 not re-opened. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
miladaslaner
added a commit
that referenced
this pull request
Jul 22, 2026
…#90) The TestRun_DurationBound/5k_eps sustained-EPS lower floor (generated ≥ 0.85× eps·after) is not portable to constrained CI: a shared 2-core runner cannot physically push 5k EPS through generate→pipeline→send, so it lands far below the floor (ratio ~0.34) and failed the v0.4.1 release Test gate, though it passes on a fast dev box (~1.04). Remove the 5k lower-floor assertion outright (founder Option (c)); retain it only at the 1k subtest (10 req/s, trivially sustainable on 2 cores). No capability probe / measured-send-rate / skip machinery. The IB-19 stop.after overrun guard (wall ≤ after+1s AND generated ≤ 1.15× eps·after) is preserved universally at BOTH 1k and 5k — it still fails-before on 9bf3c44 (1k 3.63×/9.25s, 5k 1.32×/6.02s) and passes-after on main. A governance comment marks those assertions as the IB-19 regression guard that must not be hidden behind testing.Short(), so ci.yml and release.yml keep catching the class pre-merge. Test-only: no internal/engine product code changed; IB-19 not re-opened. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
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.
What & why (IB-21, Bug — release-blocker)
TestRun_DurationBound/5k_epsblocked thev0.4.1release: its sustained-EPS lower floor (generated ≥ 0.85× eps·after) is not portable to a 2-core GitHub Actions runner, which cannot physically push 5k EPS through generate→pipeline→send (CI ratio ~0.34 →did not sustain EPS), yet passes on a fast dev box (~1.04). This is a test-portability defect — the IB-19 engine fix is correct and unchanged.Change (founder-confirmed Option (c) — test-only)
engine_test.go:528-531) and its now-deadminGenRatioconst. No capability-probe / measured-send-rate / skip-with-log machinery.wall ≤ after+1.0sANDgenerated ≤ 1.15× eps·after) universally at BOTH 1k and 5k, tolerances unchanged.testing.Short()(bothci.yml:58andrelease.yml:35run this test under-race -short).Out of scope (unchanged):
internal/engineproduct code, IB-19, overrun tolerances,ci.yml/release.ymlflags.Evidence
be2d195pre-IB-19 engine, current test overlaid): 1k wall=9.248s / ratio 3.63×, 5k wall=6.021s / ratio 1.32× → overrun assertions FAIL (5k fails via the upper bound even with the floor removed, proving the guard is preserved).go test -race ./internal/engine -run TestRun_DurationBound -v→ 1k ratio 1.10, 5k ratio 1.03, both PASS.go test -race -short ./...(==ci.yml/release.yml): 28 packages ok, 0 FAIL, exit 0.go build ./...✓ ·go vet ./...✓ ·gofmt -l .clean ·golangci-lint run ./internal/engine/...→ 0 issues.(The non-
-shortA1a/A1b throughput acceptance benchmarks inbench_a1_test.gofail on constrained/loaded hardware by design — they need ~100k EPS on a fast dedicated box and are-short-skipped in every CI/release gate. That file is untouched here.)Diff:
internal/engine/engine_test.goonly (test-only).🤖 Generated with Claude Code