fix(ask): arm the wizard_ask timeout per question, not per request - #1368
Merged
Merged
Conversation
🧙 Wizard CIRun the Wizard CI and test your changes against wizard-workbench example apps by replying with a GitHub comment using one of the following commands: Test all apps:
Test all apps in a directory:
Test an individual app:
Show more apps
Test against a Context Mill branch:
Add Results will be posted here when complete. |
Rebase the timeout fix onto the current main branch while preserving the TUI layer move.
Gilbert09
force-pushed
the
posthog/ask-timeout-per-question
branch
from
October 1, 2026 10:29
5a9041e to
0138950
Compare
Member
Author
|
The branch is up to date with 🦉 via talyn.dev |
johncwaters
reviewed
Oct 2, 2026
johncwaters
left a comment
johncwaters
left a comment
Contributor
There was a problem hiding this comment.
Note
Automated review. Not written by a human.
| private _resolveManualAuthCode: ((code: string) => void) | null = null; | ||
|
|
||
| /** Resolves the in-flight wizard_ask request. */ | ||
| private _noteAskProgress: (() => void) | null = null; |
Contributor
There was a problem hiding this comment.
Note
Automated review. Not written by a human.
Nit: the doc comment "Resolves the in-flight wizard_ask request." now sits above _noteAskProgress instead of _resolvePendingQuestion. Move the new field above that comment or give it its own one-line comment.
johncwaters
approved these changes
Oct 2, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
The seeded data-source step is the one part of a run that stops to collect live credentials, and the connection forms behind it are long: a database source asks for most of host, port, database, user, password, schema and TLS/tunnel settings, and
wizard_askcarries up to twelve questions in one call for exactly that reason.The ask overlay walks those questions one at a time and resolves the request once, at the end of the walk. The ask bridge, meanwhile, documents a per-question timeout but armed a single timer when the request opened and raced it against the whole request. So a one-line question and a nine-field connection form got the same allowance, and when it expired on a user who was still working through the form:
Telemetry for this flow shows the losses concentrated on exactly the long forms: prompts carrying six or more questions are answered far less often than one-question prompts, and they account for a disproportionate share of timed-out asks, while short prompts are answered readily. Nothing in the data distinguishes a user who walked away from one who was cut off mid-form, because nothing recorded how far they got.
Why: a user who already agreed to connect a source, and is part-way through typing its credentials, should not lose the whole source to a clock that was never meant to measure the form.
Changes
createWizardAskBridgenow hands the host anonAnswercallback alongside the question's abort signal and re-arms its timer each time it fires. A request nobody touches still expires after the same wait, so single-question prompts — the large majority — behave exactly as before.WizardAskScreencallsstore.noteAskProgress()as it advances to the next question;WizardStore.requestQuestiontakes the callback and drops it when the request settles.AgentInteraction.askandWizardUI.requestQuestioncarry it through; hosts that answer in one shot can ignore it.wizard_ask cancelledcarriesquestions_answered. A count, never a value, so a request nobody touched is distinguishable from one abandoned part-filled — which is what makes this change measurable.Test plan
vitest run— 3437 tests pass across 204 files.tsc --noEmit,eslintandprettier --checkclean on the touched files.Created with PostHog Desktop
🤖 Generated with Claude Code