Skip to content

chore: move the gateways onto fold v1.14.0 - #49

Merged
blakebauman merged 1 commit into
mainfrom
fold-v1-14-0
Aug 26, 2026
Merged

blakebauman merged 1 commit into
mainfrom
fold-v1-14-0

Conversation

@blakebauman

Copy link
Copy Markdown
Member

fold v1.14.0 shipped on 2026-08-15 ("interactive apps survive the trip through the gateway") and demo.fold.run was still serving v1.13.0 — the surface whose whole job is to be the current release, a release behind.

The bump

File Change
apps/demo/Dockerfile fold:v1.13.0 → v1.14.0
apps/enterprise/Dockerfile fold:v1.13.0 → v1.14.0
apps/uptime/src/index.ts both expectVersion entries → v1.14.0
apps/docs/.../conformance.md receipt repointed at the v1.14.0 release run

The pins and expectVersion move together because CI fails them if they drift — I ran that guard locally before pushing. The conformance receipt now points at run 31889072914, job 95022525328, resolved and confirmed conclusion: success through the API rather than by editing the old URL and hoping.

A copy claim that had drifted

v1.14.0 also vendors a new console, and the site makes two claims about it — so I opened it in a browser instead of assuming. One was wrong:

  • Map view labels each route with that upstream's live latency, over a footer reading 3 of 3 connected · every route crosses auth, policy and audit
  • Breaker state lives in the upstreams table, as its own column beside CONNECTED and LATENCY

The homepage and the walkthrough both said the map carried breaker state. Both now describe what the console actually renders. The #/upstreams?view=map link itself still resolves — the route survived the rewrite, which was the thing most at risk.

Verified on the deployed v1.14.0

  • both gateways report v1.14.0
  • tools/list → 8 tools across cfdocs (2), git (5), jobs (1), no run.fold/partialFailure
  • a minted task still resolves via tasks/get from a session that never saw it minted
  • the enterprise 401 still carries its RFC 9728 challenge
  • the map view loads and draws the three-upstream fan-out

Site and docs are deployed; both gateways are on v1.14.0.

Noted, not fixed here — a dead link inside the console

The vendored console's footer links to https://fold.run/status/, which is 404 since #44 removed that page. That link is baked into the fold binary, so it is an upstream fix in fold-run/fold, not something this repo can change. It resolves on its own if the status page comes back with fold cloud.

🤖 Generated with Claude Code

v1.14.0 shipped on 2026-08-15 and the demo was still serving v1.13.0 —
the surface whose whole job is to be the current release, a release
behind. Both Dockerfiles and both expectVersion entries move together,
which is what the CI guard exists to enforce, and the conformance
receipt repoints at the v1.14.0 release run (run 31889072914, job
95022525328 — confirmed green via the API rather than by editing the URL
and hoping).

v1.14.0 also vendors a new console, so the two console claims on the
site were re-checked against it in a browser rather than assumed. One
had drifted: the map view now labels each route with that upstream's
live latency and summarises health as "3 of 3 connected", while breaker
state lives in the upstreams table as its own column. The homepage and
the walkthrough said the map carried breaker state. Both now say what
the console actually renders.

Verified on the deployed v1.14.0: both gateways report v1.14.0, the
federation returns 8 tools across three namespaces with no
partialFailure, a minted task still resolves from a session that never
saw it minted, the enterprise 401 still carries its RFC 9728 challenge,
and the map view at #/upstreams?view=map still loads.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@blakebauman
blakebauman merged commit 726b734 into main Aug 26, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant