Files
gemini-cli/tools/gemini-cli-bot/metrics/scripts/throughput.ts
T
gemini-cli[bot] 07d3b4c470 # PR Description: Improve Metric Fidelity via Search-Based Sampling
## Summary
This PR improves the accuracy and reliability of the repository's health metrics by transitioning core velocity scripts from static sampling (`last: 100`) to search-based sampling with a fixed 7-day window.

## Changes
- **throughput.ts**: Transitioned to GitHub Search API and implemented a fixed 7-day denominator for throughput calculation. This eliminates artificial spikes caused by density-based calculations on biased samples.
- **latency.ts**: Transitioned to GitHub Search API with a fixed 7-day window to ensure metrics reflect current repository performance rather than historical averages.
- **user_touches.ts**: Transitioned to GitHub Search API to provide a more accurate count of maintainer vs. community interactions.
- **lessons-learned.md**: Updated Task Ledger (BT-66) and Decision Log to record the fix and its rationale.

## Rationale
The previous use of `repository(last: 100)` in GraphQL queries introduced a sampling bias toward creation dates and caused significant throughput anomalies (e.g., reporting 3,355 items/day during batch closures). Using the Search API with a fixed temporal window ensures that metrics represent a true point-in-time reflection of repository activity.

## Verification
- Code analysis confirms the use of Search API and defensive filtering for empty results.
- Fixed temporal denominator (7 days) verified to prevent reporting anomalies.
- Pattern consistency verified across all modified scripts.
2026-05-12 17:13:05 +00:00

126 lines
3.8 KiB
TypeScript

/**
* @license
* Copyright 2026 Google LLC
* SPDX-License-Identifier: Apache-2.0
*/
import { GITHUB_OWNER, GITHUB_REPO } from '../types.js';
import { execSync } from 'node:child_process';
try {
const now = new Date();
const sevenDaysAgo = new Date(now.getTime() - 7 * 24 * 60 * 60 * 1000);
const since = sevenDaysAgo.toISOString().split('T')[0];
const query = `
query($prQuery: String!, $issueQuery: String!) {
prSearch: search(query: $prQuery, type: ISSUE, first: 100) {
nodes {
... on PullRequest {
authorAssociation
mergedAt
}
}
}
issueSearch: search(query: $issueQuery, type: ISSUE, first: 100) {
nodes {
... on Issue {
authorAssociation
closedAt
}
}
}
}
`;
const prQuery = `repo:${GITHUB_OWNER}/${GITHUB_REPO} is:pr is:merged merged:>=${since}`;
const issueQuery = `repo:${GITHUB_OWNER}/${GITHUB_REPO} is:issue is:closed closed:>=${since}`;
const output = execSync(
`gh api graphql -F prQuery='${prQuery}' -F issueQuery='${issueQuery}' -f query='${query}'`,
{ encoding: 'utf-8' },
);
const data = JSON.parse(output).data;
const prs = (data?.prSearch?.nodes || [])
.filter((p: any) => p && p.mergedAt)
.map((p: any) => ({
association: p.authorAssociation,
date: new Date(p.mergedAt).getTime(),
}));
const issues = (data?.issueSearch?.nodes || [])
.filter((i: any) => i && i.closedAt)
.map((i: any) => ({
association: i.authorAssociation,
date: new Date(i.closedAt).getTime(),
}));
const isMaintainer = (assoc: string) =>
['MEMBER', 'OWNER', 'COLLABORATOR'].includes(assoc);
// We use a fixed 7-day denominator to prevent throughput spikes from small samples
const calculateThroughput = (
items: { association: string; date: number }[],
) => {
return items.length / 7; // items per day over 7 days
};
const prOverall = calculateThroughput(prs);
const prMaintainers = calculateThroughput(
prs.filter((i: { association: string; date: number }) =>
isMaintainer(i.association),
),
);
const prCommunity = calculateThroughput(
prs.filter(
(i: { association: string; date: number }) =>
!isMaintainer(i.association),
),
);
const issueOverall = calculateThroughput(issues);
const issueMaintainers = calculateThroughput(
issues.filter((i: { association: string; date: number }) =>
isMaintainer(i.association),
),
);
const issueCommunity = calculateThroughput(
issues.filter(
(i: { association: string; date: number }) =>
!isMaintainer(i.association),
),
);
process.stdout.write(
`throughput_pr_overall_per_day,${Math.round(prOverall * 100) / 100}\n`,
);
process.stdout.write(
`throughput_pr_maintainers_per_day,${Math.round(prMaintainers * 100) / 100}\n`,
);
process.stdout.write(
`throughput_pr_community_per_day,${Math.round(prCommunity * 100) / 100}\n`,
);
process.stdout.write(
`throughput_issue_overall_per_day,${Math.round(issueOverall * 100) / 100}\n`,
);
process.stdout.write(
`throughput_issue_maintainers_per_day,${Math.round(issueMaintainers * 100) / 100}\n`,
);
process.stdout.write(
`throughput_issue_community_per_day,${Math.round(issueCommunity * 100) / 100}\n`,
);
process.stdout.write(
`throughput_issue_overall_days_per_issue,${issueOverall > 0 ? Math.round((1 / issueOverall) * 100) / 100 : 0}\n`,
);
process.stdout.write(
`throughput_issue_maintainers_days_per_issue,${issueMaintainers > 0 ? Math.round((1 / issueMaintainers) * 100) / 100 : 0}\n`,
);
process.stdout.write(
`throughput_issue_community_days_per_issue,${issueCommunity > 0 ? Math.round((1 / issueCommunity) * 100) / 100 : 0}\n`,
);
} catch (err) {
process.stderr.write(err instanceof Error ? err.message : String(err));
process.exit(1);
}