fix(pi): prove packaged workers end to end
This commit is contained in:
@@ -140,8 +140,10 @@ async function runProductProofSamples(artifact, projectRoot, samples) {
|
||||
|
||||
function summarizeManagedMilestone(reports, cold, milestone) {
|
||||
return summarizeMeasurements(reports.flatMap(({ extension }) => (
|
||||
extension.managedWorkerMilestones
|
||||
.filter((event) => event.cold === cold && event.milestone === milestone)
|
||||
extension.managedTurns
|
||||
.filter((turn) => turn.cold === cold)
|
||||
.flatMap((turn) => turn.milestones)
|
||||
.filter((event) => event.milestone === milestone)
|
||||
.map(({ durationMs }) => durationMs)
|
||||
)));
|
||||
}
|
||||
@@ -180,32 +182,29 @@ export async function runPiReleasePerformance(options) {
|
||||
options.projectRoot,
|
||||
options.samples,
|
||||
);
|
||||
const providerMetric = (temperature, field) => summarizeMeasurements(
|
||||
providerContracts.protocols.flatMap(
|
||||
({ qualification }) => qualification[temperature][field],
|
||||
),
|
||||
);
|
||||
const managedMilestones = {
|
||||
definition: 'Actual createPiManagedWorkerOpener execution from final app.asar Main against final pi-runtime; cold creates a session, warm reopens it.',
|
||||
definition: 'Each sample is one Main-owned correlated final-ASAR turn against the final pi-runtime; cold creates a session and invokes a real packaged subagent, warm reopens the same session. No milestone is joined from an external probe.',
|
||||
cold: {
|
||||
workerQueueWaitMs: summarizeManagedMilestone(productProofs, true, 'worker.queue_wait'),
|
||||
resourcesReadyMs: summarizeManagedMilestone(productProofs, true, 'resources.ready'),
|
||||
workerSpawnMs: summarizeManagedMilestone(productProofs, true, 'worker.spawn'),
|
||||
rpcReadyMs: summarizeManagedMilestone(productProofs, true, 'rpc.ready'),
|
||||
sessionOpenMs: summarizeManagedMilestone(productProofs, true, 'session.open'),
|
||||
promptAcceptedMs: providerMetric('cold', 'promptAcceptedSamplesMs'),
|
||||
agentStartMs: providerMetric('cold', 'agentStartSamplesMs'),
|
||||
providerFirstEventMs: providerMetric('cold', 'providerFirstEventSamplesMs'),
|
||||
agentSettledMs: providerMetric('cold', 'agentSettledSamplesMs'),
|
||||
promptAcceptedMs: summarizeManagedMilestone(productProofs, true, 'prompt.accepted'),
|
||||
agentStartMs: summarizeManagedMilestone(productProofs, true, 'agent.start'),
|
||||
providerFirstEventMs: summarizeManagedMilestone(productProofs, true, 'provider.first_event'),
|
||||
agentSettledMs: summarizeManagedMilestone(productProofs, true, 'agent.settled'),
|
||||
},
|
||||
warm: {
|
||||
workerQueueWaitMs: summarizeManagedMilestone(productProofs, false, 'worker.queue_wait'),
|
||||
resourcesReadyMs: summarizeManagedMilestone(productProofs, false, 'resources.ready'),
|
||||
workerSpawnMs: summarizeManagedMilestone(productProofs, false, 'worker.spawn'),
|
||||
rpcReadyMs: summarizeManagedMilestone(productProofs, false, 'rpc.ready'),
|
||||
sessionOpenMs: summarizeManagedMilestone(productProofs, false, 'session.open'),
|
||||
promptAcceptedMs: providerMetric('warm', 'promptAcceptedSamplesMs'),
|
||||
agentStartMs: providerMetric('warm', 'agentStartSamplesMs'),
|
||||
providerFirstEventMs: providerMetric('warm', 'providerFirstEventSamplesMs'),
|
||||
agentSettledMs: providerMetric('warm', 'agentSettledSamplesMs'),
|
||||
promptAcceptedMs: summarizeManagedMilestone(productProofs, false, 'prompt.accepted'),
|
||||
agentStartMs: summarizeManagedMilestone(productProofs, false, 'agent.start'),
|
||||
providerFirstEventMs: summarizeManagedMilestone(productProofs, false, 'provider.first_event'),
|
||||
agentSettledMs: summarizeManagedMilestone(productProofs, false, 'agent.settled'),
|
||||
},
|
||||
};
|
||||
const pressureUiInteractiveMs = summarizeMeasurements(
|
||||
@@ -230,7 +229,7 @@ export async function runPiReleasePerformance(options) {
|
||||
const git = await gitEvidence(options.projectRoot);
|
||||
const scenarios = [
|
||||
{ id: 1, name: 'fresh userData metadata and Composer', evidence: ['metadata fragments', 'Electron Composer samples'], result: 'pass' },
|
||||
{ id: 2, name: 'first prompt milestone split', evidence: ['final-ASAR resources.ready/worker.spawn/rpc.ready/session.open', 'separate cold prompt.accepted/agent.start/provider.first_event/agent.settled', 'Renderer commit'], result: 'pass' },
|
||||
{ id: 2, name: 'first prompt milestone split', evidence: ['single correlated final-ASAR worker.queue_wait/resources.ready/worker.spawn/rpc.ready/session.open/prompt.accepted/agent.start/provider.first_event/agent.settled timeline per turn', 'Renderer commit'], result: 'pass' },
|
||||
{ id: 3, name: 'warm Conversation restore', evidence: ['packaged session stop/reopen/get_entries', 'warm rpc.ready samples'], result: 'pass' },
|
||||
{ id: 4, name: 'two projects provider-shaped overlap and abort isolation', evidence: providerContracts.protocols.map(({ protocol }) => protocol), result: 'pass' },
|
||||
{ id: 5, name: 'same-project read-only concurrency', evidence: ['pi-worker-pool.test.ts', 'pi-subagent.test.ts'], result: 'pass' },
|
||||
@@ -302,8 +301,11 @@ export async function runPiReleasePerformance(options) {
|
||||
budgets,
|
||||
result: 'pass',
|
||||
crossPlatformReleaseReady: false,
|
||||
releaseBlockers: [
|
||||
'macOS x64 and macOS arm64 PI-150 performance/final-artifact validation was skipped by explicit user direction',
|
||||
releaseBlockers: platform() === 'linux' && !release().toLowerCase().includes('microsoft')
|
||||
? []
|
||||
: ['Native non-WSL Linux desktop/compositor PI-150 evidence is not established by this report'],
|
||||
deferredToPi160: [
|
||||
'macOS x64 and macOS arm64 final release validation was skipped by explicit user direction and remains Not Pass',
|
||||
],
|
||||
};
|
||||
if (options.reportPath) {
|
||||
|
||||
Reference in New Issue
Block a user