Node / reticulum_mirror / MeshChatX / files / .github / workflows / bench.yml

Displaying Raw • Download

T8b949e# Benchmarks and integrity checks (push to main branches and workflow_dispatch).

T8b949e# The suite runs 3 times; reported values are median-of-medians. A smart gate

T8b949e# (noise floor + absolute delta + adaptive ratio) decides regressions — flat

T8b949e# ratio alerts on sub-ms SQLite ops are useless on shared runners.

T8b949e#

T8b949e# Pinned first-party actions (bump tag and SHA together when upgrading):

T8b949e# actions/checkout@v6.0.1 8e8c483db84b4bee98b60c0593521ed34d9990e8

T8b949e# actions/cache@v5.0.4 668228422ae6a00e4ad889ee87cd7109ec5666a7

T8b949e# benchmark-action/github-action-benchmark@v1.22.0

T8b949e# a60cea5bc7b49e15c1f58f411161f99e0df48372

Tff7b72nameTb4b4b4: Ta5d6ffBenchmarks

Tff7b72onTb4b4b4:

Tff7b72workflow_dispatchTb4b4b4:

Tff7b72pushTb4b4b4:

Tff7b72branchesTb4b4b4:

Tb4b4b4- Ta5d6ffmaster

Tb4b4b4- Ta5d6ffdev

Tff7b72concurrencyTb4b4b4:

Tff7b72groupTb4b4b4: Ta5d6ffbench-${{Ta5d6ff Ta5d6ffgithub.workflowTa5d6ff Ta5d6ff}}-${{Ta5d6ff Ta5d6ffgithub.refTa5d6ff Ta5d6ff}}

Tff7b72cancel-in-progressTb4b4b4: Ta5d6fftrue

Tff7b72permissionsTb4b4b4:

Tff7b72contentsTb4b4b4: Ta5d6ffwrite

Tff7b72pull-requestsTb4b4b4: Ta5d6ffwrite

Tff7b72envTb4b4b4:

Tff7b72FORCE_JAVASCRIPT_ACTIONS_TO_NODE24Tb4b4b4: Ta5d6fftrue

Tff7b72PYTHON_VERSIONTb4b4b4: Ta5d6ff"Ta5d6ff3.14Ta5d6ff"

Tff7b72NODE_VERSIONTb4b4b4: Ta5d6ff"Ta5d6ff24Ta5d6ff"

Tff7b72UV_VERSIONTb4b4b4: Ta5d6ff"Ta5d6ff0.11.15Ta5d6ff"

Tff7b72PNPM_VERSIONTb4b4b4: Ta5d6ff"Ta5d6ff11.1.2Ta5d6ff"

Tff7b72jobsTb4b4b4:

Tff7b72benchTb4b4b4:

Tff7b72runs-onTb4b4b4: Ta5d6ffubuntu-latest

Tff7b72timeout-minutesTb4b4b4: Ta5d6ff60

Tff7b72stepsTb4b4b4:

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffCheckout

Tff7b72usesTb4b4b4: Ta5d6ffactions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffSetTa5d6ff Ta5d6ffupTa5d6ff Ta5d6ffdevelopmentTa5d6ff Ta5d6ffenvironment

Tff7b72usesTb4b4b4: Ta5d6ff./.github/actions/setup-dev-environment

Tff7b72withTb4b4b4:

Tff7b72python-versionTb4b4b4: Ta5d6ff${{Ta5d6ff Ta5d6ffenv.PYTHON_VERSIONTa5d6ff Ta5d6ff}}

Tff7b72uv-versionTb4b4b4: Ta5d6ff${{Ta5d6ff Ta5d6ffenv.UV_VERSIONTa5d6ff Ta5d6ff}}

Tff7b72node-versionTb4b4b4: Ta5d6ff${{Ta5d6ff Ta5d6ffenv.NODE_VERSIONTa5d6ff Ta5d6ff}}

Tff7b72pnpm-versionTb4b4b4: Ta5d6ff${{Ta5d6ff Ta5d6ffenv.PNPM_VERSIONTa5d6ff Ta5d6ff}}

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffRestoreTa5d6ff Ta5d6ffbenchmarkTa5d6ff Ta5d6ffbaselineTa5d6ff Ta5d6ffcache

Tff7b72usesTb4b4b4: Ta5d6ffactions/cache@668228422ae6a00e4ad889ee87cd7109ec5666a7

Tff7b72withTb4b4b4:

Tff7b72pathTb4b4b4: Ta5d6ff./cache

Tff7b72keyTb4b4b4: Ta5d6ff${{Ta5d6ff Ta5d6ffrunner.osTa5d6ff Ta5d6ff}}-bench-baseline-v2-${{Ta5d6ff Ta5d6ffgithub.ref_nameTa5d6ff Ta5d6ff}}-${{Ta5d6ff Ta5d6ffgithub.shaTa5d6ff Ta5d6ff}}

Tff7b72restore-keysTb4b4b4: Tb4b4b4|

Tff7b72${{ runner.os }}-bench-baseline-v2-${{ github.ref_name }}-

Tff7b72${{ runner.os }}-bench-baseline-v2-

Tff7b72${{ runner.os }}-bench-baseline-

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffRunTa5d6ff Ta5d6ffbenchmarksTa5d6ff Ta5d6ff(3Ta5d6ff Ta5d6ffsuiteTa5d6ff Ta5d6ffruns,Ta5d6ff Ta5d6ffmedianTa5d6ff Ta5d6ffofTa5d6ff Ta5d6ffmedians)

Tff7b72runTb4b4b4: Tb4b4b4|

Tff7b72set -euo pipefail

Tff7b72uv run python tests/backend/run_comprehensive_benchmarks.py \

Tff7b72--runs 3 \

Tff7b72--json-output bench_results.json 2>&1 | tee bench_results.txt

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffRunTa5d6ff Ta5d6ffintegrityTa5d6ff Ta5d6fftests

Tff7b72runTb4b4b4: Tb4b4b4|

Tff7b72set -euo pipefail

Tff7b72task test:integrity 2>&1 | tee -a bench_results.txt

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffSmartTa5d6ff Ta5d6ffregressionTa5d6ff Ta5d6ffgate

Tff7b72idTb4b4b4: Ta5d6ffgate

Tff7b72runTb4b4b4: Tb4b4b4|

Tff7b72set -euo pipefail

Tff7b72mkdir -p cache

Tff7b72set +e

Tff7b72uv run python tests/backend/compare_benchmarks.py \

Tff7b72--current bench_results.json \

Tff7b72--previous ./cache/benchmark-data.json \

Tff7b72--summary bench_gate_summary.txt \

Tff7b72--noise-floor-ms 0.5 \

Tff7b72--min-abs-delta-ms 1.5

Tff7b72code=$?

Tff7b72set -e

Tff7b72echo "exit_code=${code}" >> "$GITHUB_OUTPUT"

Tff7b72if [ "${code}" != "0" ]; then

Tff7b72echo "::error::Benchmark gate reported actionable regressions (see summary)"

Tff7b72fi

Tff7b72exit "${code}"

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffUpdateTa5d6ff Ta5d6ffbenchmarkTa5d6ff Ta5d6ffbaselineTa5d6ff Ta5d6ffcache

Tff7b72ifTb4b4b4: Ta5d6ffsuccess()

Tff7b72runTb4b4b4: Tb4b4b4|

Tff7b72set -euo pipefail

Tff7b72uv run python tests/backend/compare_benchmarks.py \

Tff7b72--current bench_results.json \

Tff7b72--previous ./cache/benchmark-data.json \

Tff7b72--baseline-out ./cache/benchmark-data.json \

Tff7b72--update-baseline

Tb4b4b4- Tff7b72nameTb4b4b4: Ta5d6ffPublishTa5d6ff Ta5d6ffbenchmarkTa5d6ff Ta5d6ffchartTa5d6ff Ta5d6ff(informational)

Tff7b72ifTb4b4b4: Ta5d6ffsuccess()

Tff7b72continue-on-errorTb4b4b4: Ta5d6fftrue

Tff7b72usesTb4b4b4: Ta5d6ffbenchmark-action/github-action-benchmark@a60cea5bc7b49e15c1f58f411161f99e0df48372

Tff7b72withTb4b4b4:

Tff7b72nameTb4b4b4: Ta5d6ffMeshChatXTa5d6ff Ta5d6ffBackendTa5d6ff Ta5d6ffBenchmarks

Tff7b72toolTb4b4b4: Ta5d6ffcustomSmallerIsBetter

Tff7b72output-file-pathTb4b4b4: Ta5d6ffbench_results.json

Tff7b72external-data-json-pathTb4b4b4: Ta5d6ff./cache/benchmark-data.json

Tff7b72github-tokenTb4b4b4: Ta5d6ff${{Ta5d6ff Ta5d6ffsecrets.GITHUB_TOKENTa5d6ff Ta5d6ff}}

T8b949e# Chart/comment only — smart gate is the fail path.

Tff7b72alert-thresholdTb4b4b4: Ta5d6ff"Ta5d6ff1000%Ta5d6ff"

Tff7b72fail-on-alertTb4b4b4: Ta5d6fffalse

Tff7b72comment-on-alertTb4b4b4: Ta5d6fffalse

Tff7b72summary-alwaysTb4b4b4: Ta5d6fftrue