Files
gemini-cli/tools/gemini-cli-bot/metrics/scripts/backlog_age.ts
T
gemini-cli[bot] 3a06655ec5 ### Backlog Health & Stale Policy Optimization
#### Problem Statement
Current repository metrics (`latency`, `throughput`) suffer from **survivorship bias**: they only sample the last 100 *closed* items, making the repository appear healthier than it is. Meanwhile, a stable backlog of **2342 open issues** and **442 open PRs** persists, largely due to "staleness immunity" for `help wanted` items and throttling in the standard stale workflow.

#### Changes
1.  **New Metric: Backlog Age**: Added `tools/gemini-cli-bot/metrics/scripts/backlog_age.ts` to measure the median age of the oldest 100 open issues and PRs. This exposes the "Slow Path" bottleneck that was previously invisible.
2.  **Stale Policy Throttling Fix**: Increased `operations-per-run` from 30 (default) to 200 in `.github/workflows/stale.yml` to allow the daily cron to actually make progress on the large backlog.
3.  **Help-Wanted Expiration**: Updated `gemini-scheduled-stale-issue-closer.yml` to remove the infinite exemption for `help wanted` issues. They are now eligible for stale closure if they are older than 180 days and have no recent human activity.

#### Expected Impact
- **Visibility**: The new `backlog_age` metrics will likely show high values initially, providing a baseline for backlog reduction efforts.
- **Efficiency**: Throttling fix will increase the rate of stale item closure.
- **Backlog Reduction**: The 6-month expiration for `help wanted` will finally address legacy "immortal" issues that have been bloating the backlog for years.

This is a surgical PR focused on repository health and metric accuracy.
2026-05-01 00:14:12 +00:00

65 lines
2.1 KiB
TypeScript

/**
* @license
* Copyright 2026 Google LLC
* SPDX-License-Identifier: Apache-2.0
*/
import { execSync } from 'node:child_process';
import { GITHUB_OWNER, GITHUB_REPO } from '../types.js';
try {
const query = `
query($owner: String!, $repo: String!) {
repository(owner: $owner, name: $repo) {
issues(states: OPEN, first: 100, orderBy: {field: CREATED_AT, direction: ASC}) {
totalCount
nodes {
createdAt
}
}
pullRequests(states: OPEN, first: 100, orderBy: {field: CREATED_AT, direction: ASC}) {
totalCount
nodes {
createdAt
}
}
}
}
`;
const output = execSync(
`gh api graphql -F owner=${GITHUB_OWNER} -F repo=${GITHUB_REPO} -f query='${query}'`,
{ encoding: 'utf-8', stdio: ['ignore', 'pipe', 'ignore'] },
).trim();
const parsed = JSON.parse(output);
const data = parsed?.data?.repository;
if (data?.issues?.totalCount > 100) {
process.stderr.write(`Warning: Backlog has ${data.issues.totalCount} issues, but only the oldest 100 were used for median calculation.\n`);
}
if (data?.pullRequests?.totalCount > 100) {
process.stderr.write(`Warning: Backlog has ${data.pullRequests.totalCount} PRs, but only the oldest 100 were used for median calculation.\n`);
}
const calculateMedianAgeDays = (nodes: { createdAt: string }[]) => {
if (!nodes || nodes.length === 0) return 0;
const now = Date.now();
const ages = nodes.map(
(n) => (now - new Date(n.createdAt).getTime()) / (1000 * 60 * 60 * 24),
);
ages.sort((a, b) => a - b);
const mid = Math.floor(ages.length / 2);
return ages.length % 2 !== 0
? ages[mid]
: (ages[mid - 1] + ages[mid]) / 2;
};
const issueAge = calculateMedianAgeDays(data?.issues?.nodes ?? []);
const prAge = calculateMedianAgeDays(data?.pullRequests?.nodes ?? []);
process.stdout.write(`backlog_age_issue_median_days,${Math.round(issueAge * 100) / 100}\n`);
process.stdout.write(`backlog_age_pr_median_days,${Math.round(prAge * 100) / 100}\n`);
} catch (err) {
process.stderr.write(err instanceof Error ? err.message : String(err));
process.exit(1);
}