In the nodejs/citgm repo there is a backlog of issues raised, PRs submitted, and PRs that have landed, but haven't been released:
- citgm supports "node": "20.x || 22.x || 24.x" missing out Node.js 26.x. Resolved in chore: update supported Node.js versions (22, 24 & 26) citgm#1116 on Aug 10, 2026. Unreleased.
- Multiple packages sourced from archived repos are failing (JSONStream, leveldown, q, gulp-util & node-sqlite3). Archived repos are by definition unsupported, since it is not possible to report issues or for updates to be released. The removal of q was actioned, however the removal is unreleased. See issue Archived (unsupported) packages - JSONStream, leveldown & q etc. citgm#1113.
- Multiple packages (@babel/core, clinic, fastify, jest, node-gyp, pino & uuid) are failing in all current variants. See issue Review all tests citgm#1136.
- Two additional failures occur when testing under npm@12. See npm 12 - npm run tap (2 failures) citgm#1130.
The list of packages in nodejs/citgm/lib/lookup.json should contain only packages that successfully test in citgm and that are still supported. Any particular variants that do not pass should include skip entries.
Impact
The Jenkins Node.js-citgm jobs are testing packages that always fail. This unnecessarily increases the test duration (last citgm-smoker was 5 hr 57 min). The normal state for citgm-smoker is that it always fails, instead of only failing when there is a regression.
Reactions are currently unavailable
In the nodejs/citgm repo there is a backlog of issues raised, PRs submitted, and PRs that have landed, but haven't been released:
The list of packages in nodejs/citgm/lib/lookup.json should contain only packages that successfully test in citgm and that are still supported. Any particular variants that do not pass should include skip entries.
Impact
The Jenkins Node.js-citgm jobs are testing packages that always fail. This unnecessarily increases the test duration (last citgm-smoker was 5 hr 57 min). The normal state for citgm-smoker is that it always fails, instead of only failing when there is a regression.