Files
fusion/packages/dashboard/app/utils/taskStatusBadgeLabel.ts
gsxdsm 1f5c44ac71 FN-8826: restore WIP lifecycle badges
Restore visible lifecycle badges for empty-status tasks in WIP lanes.

- Derive fallback badge labels from resolved workflow WIP traits and lane names
- Render the fallback consistently on board cards and desktop/mobile list rows
- Cover empty, custom-lane, populated, and paused status precedence

Files changed:
 .changeset/fn-8826-dashboard-wip-badge.md          |  7 +++
 packages/dashboard/app/components/ListView.tsx     | 37 +++++++++++--
 packages/dashboard/app/components/TaskCard.tsx     | 19 +++++--
 .../app/components/__tests__/ListView.test.tsx     | 61 ++++++++++++++++++++++
 .../app/components/__tests__/TaskCard.test.tsx     | 38 +++++++++++++-
 .../dashboard/app/utils/taskStatusBadgeLabel.ts    | 24 +++++++++
 6 files changed, 177 insertions(+), 9 deletions(-)

Fusion-Task-Id: FN-8826

Fusion-Task-Lineage: fd9c857a-a25f-44d0-ae66-502bc1bd860e

Co-authored-by: Fusion (runfusion.ai) <noreply@runfusion.ai>
2026-08-07 20:33:42 -07:00

157 lines
7.5 KiB
TypeScript

/*
FNXC:MergeQueue 2026-07-15-10:45:
AI merge sets task.status to reviewing/landing for most of the live merge window. Board/list badges must never show those raw engine strings; map the full active-merge pipeline to operator-facing Merging… (and Merging fixes… for merging-fix).
*/
import type { TFunction } from "i18next";
import type { Task } from "@fusion/core";
import { isActiveMergeStatus } from "../../../core/src/merge/active-merge-status";
/*
FNXC:TaskStatusBadge 2026-07-22-00:00:
FN-8475 restores truthful status visibility: Coding (Ideas) deliberately plans in Todo,
so a real non-queued task status must not be hidden based only on its board column.
Queued remains an intake-only presentation exclusion at TaskCard and ListView call sites.
*/
export function hasTaskStatusBadge(status: string | null | undefined): boolean {
return typeof status === "string" && status.trim().length > 0;
}
export interface TaskWipLifecycleBadgeContext {
/** True when the task's resolved workflow column occupies an implementation/WIP slot. */
isWipColumn: boolean;
/** The resolved workflow column name, when board metadata has loaded. */
lifecycleLabel?: string | null;
}
/*
FNXC:TaskStatusBadge 2026-08-08-03:02:
A WIP lifecycle role supplies a badge only when the persisted status is empty: engine status,
pause, workflow-gate, and other call-site precedence remains authoritative. Resolve WIP from
workflow traits before calling this helper so renamed/custom implementation lanes receive their
own lifecycle label; metadata gaps retain the safe legacy "In progress" fallback without writing
that presentation state back to task.status.
*/
export function getTaskWipLifecycleBadgeLabel(
status: string | null | undefined,
t: TFunction<"app">,
{ isWipColumn, lifecycleLabel }: TaskWipLifecycleBadgeContext,
): string | null {
if (!isWipColumn || hasTaskStatusBadge(status)) return null;
return lifecycleLabel?.trim() || t("tasks.statusInProgress", "In progress");
}
/*
FNXC:TaskStatusConsistency 2026-08-05-03:49:
A planner agent-log event can reach the shared board snapshot before the status update that changes
`needs-replan` to `planning`. Its `recentAgentActivityAt` is written only after the log proves it is
newer than the row, so this is explicit live-planner evidence rather than arrival-order guessing.
Use it only for the parked-revision token while the event is inside the five-minute live window and
the engine is not globally paused: undefined status remains unknown and an idle needs-replan card
remains Queued to revise. TaskCard, both ListView paths, and task detail share this predicate so a
modal that has already fetched planning cannot contradict the board during that short SSE gap.
FNXC:TaskStatusConsistency 2026-08-05-04:54:
A parseable historical log is not liveness. The transient `needs-replan` override expires after the
shared five-minute heartbeat window and is suppressed during global pause, returning the operator to
Queued to revise instead of leaving every affected surface stuck on Planning.
*/
export const PLANNER_ACTIVITY_LIVE_WINDOW_MS = 5 * 60_000;
export interface TaskPlanningActivityOptions {
globalPaused?: boolean;
now?: number;
}
export function isTaskPlanningActive(
task: Pick<Task, "status" | "recentAgentActivityAt">,
{ globalPaused = false, now = Date.now() }: TaskPlanningActivityOptions = {},
): boolean {
if (task.status === "planning") return true;
if (globalPaused || task.status !== "needs-replan" || typeof task.recentAgentActivityAt !== "string") return false;
const activityAt = Date.parse(task.recentAgentActivityAt);
return Number.isFinite(activityAt)
&& activityAt <= now
&& now - activityAt <= PLANNER_ACTIVITY_LIVE_WINDOW_MS;
}
/*
FNXC:TaskStatusBadge 2026-08-01-01:30 (operator: "make the badges more descriptive"):
Three waiting states read as ACTIVITY on the board and made healthy queueing look broken:
"Revising" while merely queued for a replan slot, and bare "queued" while parked behind another
task's file-scope lease (the reason lived only in a log line). The badge now names the wait:
"Queued to revise" (idle needs-replan) and "Queued behind FN-XXXX" (overlap lease). Liveness and
the overlap id are caller-supplied so this stays a pure mapper; omitted → legacy labels, so every
existing caller keeps its behavior.
*/
export interface TaskStatusBadgeContext {
/** True when no agent session is live for the card (queued, not running). */
idle?: boolean;
/** The task currently holding the overlapping file-scope lease, when queued behind one. */
overlapBlockedBy?: string | null;
}
export function getTaskStatusBadgeLabel(
status: string | null | undefined,
t: TFunction<"app">,
/*
FNXC:TaskStatusBadge 2026-07-19-02:55 (U12 / R2 / R11):
Workflow-step state wins over the raw status vocabulary. A card whose Plan Review is running
reads "Plan Review" — the step's own IR-declared name — instead of the engine token "planning"
or "needs-replan". Pass `getRunningWorkflowStepLabel(task)` here; omit it and the legacy status
mapping below is unchanged, so every existing caller keeps its behavior.
*/
workflowStepLabel?: string,
context?: TaskStatusBadgeContext,
): string {
/*
FNXC:TaskStatusBadge 2026-07-19-09:40:
Every active-merge status ("merging", "merging-pr", "merging-fix", "reviewing", "landing") must
win over a still-running workflow-step label (a pre-merge step's startedAt-without-completedAt
state can survive into the merge pipeline). Checking the status before the workflow-step override
enforces this for every caller (TaskCard, ListView grouped rows, ListView table rows) instead of
relying on per-call-site pre-checks. "merging-fix" keeps its distinct "Merging fixes…" label.
*/
if (isActiveMergeStatus(status)) {
return status === "merging-fix"
? t("tasks.statusMergingFix", "Merging fixes…")
: t("tasks.statusMerging", "Merging…");
}
if (workflowStepLabel) return workflowStepLabel;
if (!status) return "";
/*
FNXC:TaskStatusBadge 2026-07-22-09:22:
FN-8493 requires operator copy "Revising" while a revise/replan cycle is in progress. Keep the
engine token "needs-replan" unchanged and map centrally so board cards and list rows agree.
*/
if (status === "needs-replan") {
// Idle = waiting for a planning slot; the live revise cycle shows as "Revising"/step label.
return context?.idle
? t("tasks.statusReplanQueued", "Queued to revise")
: t("tasks.statusReplan", "Revising");
}
if (status === "queued" && context?.overlapBlockedBy) {
return t("tasks.statusQueuedBehind", "Queued behind {{taskId}}", { taskId: context.overlapBlockedBy });
}
/*
FNXC:TaskStatusBadge 2026-08-01-03:20 (operator: one queued family, like the planning badge):
Raw "queued" (scheduler capacity/dependency marker) previously fell through to the verbatim
lowercase engine token. Map it so every queued state reads as the same badge family.
*/
if (status === "queued") {
return t("tasks.statusQueued", "Queued");
}
/*
FNXC:TaskStatusBadge 2026-07-26-14:05:
"planning" is an engine token, not operator copy, and it was only ever hidden because U12's
workflow-step override happened to cover the same cards. Now that the Plan Review gate has its own
badge, callers suppress the step-name override while that badge renders — which uncovered the raw
lowercase token underneath. Map it to the same "Planning" copy the transient-planner badge already
uses so the two paths cannot read differently.
*/
if (status === "planning") {
return t("tasks.statusPlanning", "Planning");
}
return status;
}