fix(rivetkit/wf): replay successful result mechanism by abcxff · Pull Request #4310 · rivet-dev/rivet

abcxff · 2026-02-25T15:05:15Z

FIXES RVT-6000

Description

Please include a summary of the changes and the related issue. Please also include relevant motivation and context.

Type of change

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to not work as expected)
This change requires a documentation update

How Has This Been Tested?

Please describe the tests that you ran to verify your changes.

Checklist:

My code follows the style guidelines of this project
I have performed a self-review of my code
I have commented my code, particularly in hard-to-understand areas
I have made corresponding changes to the documentation
My changes generate no new warnings
I have added tests that prove my fix is effective or that my feature works
New and existing unit tests pass locally with my changes

railway-app · 2026-02-25T15:05:26Z

🚅 Deployed to the rivet-pr-4310 environment in rivet-frontend

Service	Status	Web	Updated (UTC)
website	😴 Sleeping (View Logs)	Web	Feb 25, 2026 at 3:19 pm
frontend-inspector	😴 Sleeping (View Logs)	Web	Feb 25, 2026 at 3:18 pm
frontend-cloud	😴 Sleeping (View Logs)	Web	Feb 25, 2026 at 3:15 pm
mcp-hub	✅ Success (View Logs)	Web	Feb 25, 2026 at 3:06 pm
ladle	❌ Build Failed (View Logs)	Web	Feb 25, 2026 at 3:06 pm

abcxff · 2026-02-25T15:05:29Z

fix(rivetkit/wf): replay successful result mechanism #4310 👈 (View in Graphite)
main

How to use the Graphite Merge Queue

Add the label merge-queue to this PR to add it to the merge queue.

You must have a Graphite account in order to use the merge queue. Sign up using this link.

_{An organization admin has enabled the Graphite Merge Queue in this repository.} _{Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.}

This stack of pull requests is managed by Graphite. Learn more about stacking.

linear · 2026-02-25T15:07:03Z

RVT-6000 rvtkit/wf: Fix void-step retry mechanism for

claude · 2026-02-25T15:10:53Z

PR Review: fix(rivetkit/wf): replay successful result mechanism

The fix correctly addresses a real bug: void steps/loops (those returning undefined) were being re-executed on workflow restart because the old replay guard stepData.output \!== undefined treats undefined outputs as "not yet run." Using metadata.status === "completed" as the authoritative source is the right fix.

What's good

Core fix is sound. metadata.status === "completed" correctly captures completion regardless of the return value.
Redundant fallback preserved. The legacy loopData.output \!== undefined check remains as a backwards-compatibility guard for loop entries persisted before metadata tracking was introduced.
Test coverage is solid. Both void-step and void-loop restart scenarios are tested independently across all driver modes.

Issues / suggestions

1. Missing debug log after refactor (minor)

The original code logged when a step was replayed from history:

this.log("debug", { msg: "replaying step from history", step: config.name, key });

This was silently dropped in the refactor. This log is useful for diagnosing replay issues. Consider restoring it inside the new guard:

if (metadata.status === "completed" || stepData.output \!== undefined) {
    this.log("debug", { msg: "replaying step from history", step: config.name, key });
    return stepData.output as T;
}

2. loadMetadata now always called on step replay (minor performance regression)

Before the fix, non-void steps had a fast path — they returned immediately without loading metadata. Now loadMetadata is always called first, adding an I/O round-trip for every historical non-void step replay too. The fix is necessary for correctness, but a comment noting the tradeoff (or an optimized path like checking output first and skipping metadata load) would help future maintainers understand why the order changed.

3. Rollback mode behavior change for completed void loops (needs confirmation)

In the original code, a completed void loop in rollback mode fell through to re-execute one iteration (because both guards evaluated false):

// rollback guard: loopData.output === undefined → no early return
// completion guard: loopData.output === undefined → no early return
// → rollbackSingleIteration = true, loop re-runs

After the fix, metadata.status === "completed" causes an early return before the loop re-executes. This is arguably more correct (a completed loop doesn't need a rollback iteration), but it's an implicit behavior change worth documenting with a comment, and ideally a test covering a void loop inside a rollback-mode workflow.

4. metadata.status = "running" reset for an already-completed loop with undefined output (subtle)

If metadata.status \!== "completed" but loopData.output \!== undefined, the code returns early at the legacy guard (line 663) — so the if (metadata) { metadata.status = "running"; ... } block is never reached. That's fine. But if both are false (loop in progress, no output yet), the code proceeds to set metadata.status = "running", which correctly restores the running state. The control flow is correct, but non-obvious — a brief comment at the `if (metadata)" block explaining "only reached when loop is resuming from an incomplete state" would help.

CI workflow change

Removal of \!@rivetkit/example-registry from the turbo build filter looks intentional (restoring a previously excluded package). Worth confirming this doesn't break the pkg-pr-new publish step if @rivetkit/example-registry has no publishable build output.

Summary

The bug fix is correct and the test coverage is appropriate. The four items above are minor (one log, one perf note, one rollback edge case needing a test, one readability comment). None are blockers, but item 3 (rollback behavior for void loops) is worth a quick sanity check.

pkg-pr-new · 2026-02-25T15:14:58Z

More templates

@rivetkit/cloudflare-workers

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/cloudflare-workers@4310

@rivetkit/framework-base

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/framework-base@4310

@rivetkit/next-js

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/next-js@4310

@rivetkit/react

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/react@4310

rivetkit

pnpm add https://pkg.pr.new/rivet-dev/rivet/rivetkit@4310

@rivetkit/sql-loader

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/sql-loader@4310

@rivetkit/sqlite-vfs

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/sqlite-vfs@4310

@rivetkit/traces

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/traces@4310

@rivetkit/workflow-engine

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/workflow-engine@4310

@rivetkit/virtual-websocket

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/virtual-websocket@4310

@rivetkit/engine-runner

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/engine-runner@4310

@rivetkit/engine-runner-protocol

pnpm add https://pkg.pr.new/rivet-dev/rivet/@rivetkit/engine-runner-protocol@4310

commit: 2841137

railway-app bot temporarily deployed to rivet-frontend / rivet-pr-4310 February 25, 2026 15:05 Destroyed

abcxff requested a review from NathanFlurry February 25, 2026 15:06

abcxff marked this pull request as ready for review February 25, 2026 15:07

fix(rivetkit/wf): replay successful result mechanism

2841137

abcxff force-pushed the 02-25-fix_rivetkit_wf_replay_successful_result_mechanism branch from d5f9531 to 2841137 Compare February 25, 2026 15:11

railway-app bot temporarily deployed to rivet-frontend / rivet-pr-4310 February 25, 2026 15:11 Destroyed

NathanFlurry approved these changes Feb 26, 2026

View reviewed changes

NathanFlurry merged commit dad1242 into main Feb 26, 2026
15 of 24 checks passed

NathanFlurry deleted the 02-25-fix_rivetkit_wf_replay_successful_result_mechanism branch February 26, 2026 09:54

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(rivetkit/wf): replay successful result mechanism#4310

fix(rivetkit/wf): replay successful result mechanism#4310
NathanFlurry merged 1 commit intomainfrom
02-25-fix_rivetkit_wf_replay_successful_result_mechanism

abcxff commented Feb 25, 2026 •

edited

Loading

Uh oh!

railway-app bot commented Feb 25, 2026 •

edited

Loading

Uh oh!

abcxff commented Feb 25, 2026

Uh oh!

linear bot commented Feb 25, 2026

Uh oh!

claude bot commented Feb 25, 2026 •

edited

Loading

Uh oh!

pkg-pr-new bot commented Feb 25, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

abcxff commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Type of change

How Has This Been Tested?

Checklist:

Uh oh!

railway-app bot commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

abcxff commented Feb 25, 2026

How to use the Graphite Merge Queue

Uh oh!

linear bot commented Feb 25, 2026

Uh oh!

claude bot commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Review: fix(rivetkit/wf): replay successful result mechanism

What's good

Issues / suggestions

CI workflow change

Summary

Uh oh!

pkg-pr-new bot commented Feb 25, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

abcxff commented Feb 25, 2026 •

edited

Loading

railway-app bot commented Feb 25, 2026 •

edited

Loading

claude bot commented Feb 25, 2026 •

edited

Loading