Files
PaperClipAI/.github
DottaandClaude Opus 5.5 0e8447fe77 fix(release): retry the finalize install while a large tarball lags its npm metadata (#21925)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - Since #21917, each canary's npm confirmation, git tag, Docker build
and onboarding smoke run in `canary-finalize.yml`
> - The first canary through it, `2026.1116.0-canary.814700`, passed the
master-position, visibility, provenance (31 of 31) and dist-tag checks
> - It then failed the clean-prefix install: the 524 MB
`@paperclipai/server` tarball returned 404 while npm already listed the
version
> - Without a tag, nightly cannot select that canary, and no Docker
image or smoke runs for it
> - This pull request retries that install for a bounded time
> - The benefit is that a large tarball's normal lag no longer stops a
canary from finishing

## Linked Issues or Issue Description

**What happened?**
Canary finalize run 43927092510 for `2026.1116.0-canary.814700`:
`npm error 404 Not Found - GET
https://registry.npmjs.org/@paperclipai/server/-/server-2026.1116.0-canary.814700.tgz`.
npm's packument records that version at 18:31:41Z. The install ran at
18:32:03Z, and the tarball still returned 404 at 18:34:39 and 18:35:15.
`@paperclipai/shared` and `paperclipai` tarballs at the same version
returned 200.

**Expected behavior**
Finalize waits, within a bound, until the published tarball downloads,
then tags the canary.

Refs #21917

## What Changed

- `scripts/release-lib.sh`: `wait_for_npm_installable` retries
`verify_npm_installable` up to N attempts with a fixed delay.
- `scripts/confirm-canary-publication.sh`: uses it with 40 attempts × 30
s (`NPM_INSTALL_VERIFY_ATTEMPTS`, `NPM_INSTALL_VERIFY_DELAY_SECONDS`).
- `.github/workflows/canary-finalize.yml`: `confirm_canary` timeout 45 →
75 min, to fit the existing 30-minute visibility budget plus the
20-minute install budget.
- Test and `doc/RELEASING.md`.

## Verification

- `node --test scripts/release-lib.test.mjs`: 23 of 23 pass, including a
new test: immediate success, recovery after 3 failures (4 attempts,
warning names attempt 3/5), and failure after the bound (4 of 4
attempts).
- `node --test scripts/__tests__/release-verify-workflow.test.mjs`: 73
of 73 pass.
- After merge: the next canary's finalize run passes the install and
pushes its tag. `canary.814700` itself is recovered by rerunning its
finalize run once its tarball downloads.

## Risks

- Low. A tarball that never appears still fails finalize, after 20
minutes instead of at once.

## Model Used

- Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, in Claude Code with
tool use and code execution.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [ ] I will address all Greptile and reviewer comments before
requesting merge

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-16 14:24:40 -06:00
..