Skip to content

[pull] main from vercel:main#454

Merged
pull[bot] merged 4 commits into
erickirt:mainfrom
vercel:main
Jul 13, 2026
Merged

[pull] main from vercel:main#454
pull[bot] merged 4 commits into
erickirt:mainfrom
vercel:main

Conversation

@pull

@pull pull Bot commented Jul 13, 2026

Copy link
Copy Markdown

See Commits and Changes for more details.


Created by pull[bot] (v2.0.0-alpha.4)

Can you help keep this open source service alive? 💖 Please sponsor : )

karthikscale3 and others added 4 commits July 13, 2026 09:58
* telemetry: emit client-observed workflow.stream.write span per flush batch

Complements the existing workflow.stream.read TTFC span: each flushed
batch emits a back-dated CLIENT span covering the app-perceived write
latency (buffer dwell + RPC), with buffer_dwell_ms / chunks / bytes
attributes so client-side batching cost (flush timer, turbo run-ready
barrier) can be told apart from network/server time. Failed flushes
keep the batch's original t0 so a retried batch reports its full dwell.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore: tighten changeset

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* telemetry: rename flush span to workflow.stream.flush; DEBUG-log world-vercel OTEL load failure

- workflow.stream.write is taken by world-vercel's per-request RPC span
  (#2857, chunk_rtt); the per-batch flush span gets its own name so the
  two stay distinguishable in trace queries. Attributes move to
  workflow.stream.flush.{buffer_dwell_ms,chunks,bytes}, operation=flush.
- world-vercel's @opentelemetry/api load failure was silently latched as
  null, which also swallows bundler/resolution failures in apps that DO
  register a tracer (observed in production: workbench apps emit core
  spans but none of world-vercel's). Log the reason under
  DEBUG=workflow:* so the failure mode is diagnosable.
- Document workflow.stream.flush in the tracing docs.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore: tighten changeset

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* fix(world): parse timezone-naive analytics timestamps as UTC

ClickHouse-backed analytics endpoints serialize DateTime64 values as
timezone-naive strings ('2026-07-13 17:09:11.593'), UTC by convention.
The analytics schemas coerced them with z.coerce.date(), i.e.
new Date(value), which interprets naive strings in the process's LOCAL
timezone. That is only correct when the process runs in UTC — the
deployed observability web app's server actions, which is why the web
UI appears unaffected — and wrong by the local UTC offset everywhere
else: the CLI on a laptop showed runs 'starting in about 7 hours'
(PDT), and 'workflow web --localUi' shares the bug.

Replace the coercion with a preprocess that normalizes naive datetime
strings to an explicit Z designator before parsing. Values already
carrying timezone info (Z or ±hh:mm) and non-string inputs (Date,
epoch) pass through unchanged.

Tests pin the contract and were verified under TZ=America/Los_Angeles
and TZ=Asia/Tokyo (3 of 4 fail against the old coercion in PDT; plain
UTC CI cannot distinguish the two, which is how this shipped).

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Signed-off-by: Nathan Rajlich <n@n8.io>

---------

Signed-off-by: Nathan Rajlich <n@n8.io>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
…#2898)

* build: declare typescript (catalog:) in every package that runs tsc

Twenty packages invoke tsc in their build/typecheck scripts without
declaring a typescript dependency, resolving whatever tsc pnpm happens
to leave reachable. That broke locally after the TypeScript 6 upgrade
(#2700): base.json now uses the TS6-only 'types': ['*'] wildcard, and
worktrees carrying pre-upgrade node_modules/.bin/tsc shims (orphaned
typescript@5.9.3 bins that pnpm never refreshes for an undeclared
dependency) fail with TS2688 'Cannot find type definition file for *'.

Declaring 'typescript': 'catalog:' (the convention nest already
follows) makes pnpm own each package's tsc bin, so version upgrades
refresh the shims and this staleness class cannot recur. Packages
without tsc in their scripts are left unchanged.

Full pnpm build: 27/27 tasks green.

* Address review: drop duplicate zod devDep; regenerate lockfile minimally

- packages/world listed zod in both dependencies and devDependencies
  (pre-existing on main, surfaced by the devDependencies sort) — keep
  the runtime dependency only.
- Regenerate pnpm-lock.yaml from a pristine main baseline with
  --lockfile-only (a clean-main run produces zero diff, so main has no
  drift). Remaining non-typescript changes are mechanical consequences
  of the change itself: typescript is an (optional) peer of several
  tooling dependencies, so declaring it in 20 importers creates new
  peer-resolution snapshot variants and prunes the now-orphaned old
  ones; plus one radix-ui 1.6.1->1.6.2 refresh in docs caused by its
  floating 'latest' specifier.
- Validated: pnpm install --frozen-lockfile succeeds; full build 27/27.
@pull pull Bot locked and limited conversation to collaborators Jul 13, 2026
@pull pull Bot added the ⤵️ pull label Jul 13, 2026
@pull pull Bot merged commit b01ed54 into erickirt:main Jul 13, 2026
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants