Files
multica/packages/views/modals/run-confirm.tsx
Jiayuan Zhang 7803a5b9ea feat(ui): establish a role-named type scale and migrate ad-hoc font sizes (MUL-5451) (#6136)
tokens.css defined colours, radii and font families but not a single --text-*
step, so font sizes had no baseline to align to and grew wherever they were
needed: 51 distinct sizes across web + desktop, 370 written as arbitrary
values, six at half a pixel (10.5 / 11.5 / 12.5 / 13.5 / 14.5 / 15.5px).
text-xs and text-sm carried nearly all UI text while the range between them —
11, 13, 15px — could only be reached with arbitrary values. Hierarchy does not
come from having more sizes; past a handful, each extra size makes the
hierarchy blurrier.

Add ten role-named steps, each with its own line-height so leading cannot
fragment the way size did, and move every product-UI call site onto them.
Steps are named for what the text is for, not for a t-shirt size, because
that is what keeps the scale from drifting again.

Six steps deliberately keep the exact size/line-height pairs of the Tailwind
defaults they replace, so the ~1,900-call-site rename moves nothing on screen.
The visible changes are confined to former arbitrary values snapping to a step:
8/9/10px -> micro (11px) on badges and overlines; 17 -> 18; 22 -> 24; 30
(text-3xl) -> 36 on headings and stat numbers; 12.8px -> label (13px) on small
buttons and toggles. Half-pixel sizes are gone.

This supersedes #6108, which was reverted by #6116 because the sidebar group
labels rendered at the inherited 16px. The cause was not the scale but cn():
`text-<x>` is ambiguous in Tailwind, and tailwind-merge resolves it against a
table listing only the default sizes, so it filed every role step under
text-colour and dropped whichever of `text-caption` /
`text-sidebar-foreground/70` came first. Registering the steps as a font-size
class group restores the real conflict groups — size beats size, colour beats
colour, the two coexist — and a test pins the list against the scale, since
the failure is silent in source.

Hand-written CSS is covered too. The transcript kept a 12.5px body long after
every Tailwind call site was on the scale, so the "no half-pixel sizes" claim
was true of the classes and false of the product; the editor's prose, code and
mermaid ramps had the same blind spot, and seven of their eight values already
equalled a step exactly. All now reference var(--text-*). The guard test reads
raw `font-size:` declarations as well as class names, exempting only the 16px
iOS input-zoom workaround in base.css and the landing pages' marketing ramp.

apps/mobile (own NativeWind config) and apps/docs (fumadocs' own type system)
keep Tailwind's default scale and are untouched. Landing display type
(rem/clamp, 2.2-6.4rem) stays on its separate ramp, as do four decorative
emoji / serif-hero sizes.

Verified on a running local stack: pinned sidebar rows and group labels
measure 12px/16px, nav items 14px/20px — identical to pre-migration. An audit
of every rendered font size across the product surfaces finds nothing off the
scale; the only exceptions are avatar initials and emoji, which
actor-avatar.tsx sizes proportionally to the avatar diameter by design.

Co-authored-by: Lambda <lambda@multica.ai>
Co-authored-by: multica-agent <github@multica.ai>
2026-07-30 13:42:33 +08:00

222 lines
9.3 KiB
TypeScript

"use client";
import { useMemo, useState, type ReactNode } from "react";
import { useQuery } from "@tanstack/react-query";
import { toast } from "sonner";
import {
Dialog,
DialogContent,
DialogDescription,
DialogFooter,
DialogHeader,
DialogTitle,
} from "@multica/ui/components/ui/dialog";
import { Button } from "@multica/ui/components/ui/button";
import { Textarea } from "@multica/ui/components/ui/textarea";
import { Spinner } from "@multica/ui/components/ui/spinner";
import type { IssueAssigneeType, UpdateIssueRequest } from "@multica/core/types";
import { useUpdateIssue, useBatchUpdateIssues } from "@multica/core/issues/mutations";
import { useActorName } from "@multica/core/workspace/hooks";
import { useWorkspaceId } from "@multica/core/hooks";
import { agentListOptions, squadListOptions } from "@multica/core/workspace/queries";
import { runtimeListOptions, readRuntimeCliVersion, handoffSupported } from "@multica/core/runtimes";
import { useT } from "../i18n";
const MAX_HANDOFF_NOTE = 2000;
// i18next inlines {{name}} into the sentence, but the actor's position varies by
// language ("{{name}} 会…" vs "Once assigned, {{name}} will…" vs "{{name}}'s
// leader…"). Fence the name with a sentinel so we can bold just that span at
// render time without splitting copy into per-language prefix/suffix keys.
const NAME_FENCE = "\u0000";
function boldName(text: string): ReactNode {
const parts = text.split(NAME_FENCE);
if (parts.length !== 3) return text;
return (
<>
{parts[0]}
<span className="font-semibold text-foreground">{parts[1]}</span>
{parts[2]}
</>
);
}
interface RunConfirmData {
issueIds?: string[];
// Assign is the only mode: agent/squad assignment is the sole issue write that
// needs the pre-trigger confirmation. Batch status changes apply directly now
// (MUL-4155), so there is no "status" mode.
mode?: "assign";
assigneeType?: IssueAssigneeType;
assigneeId?: string;
assigneeName?: string;
}
/**
* Assignment confirmation for issues that may start agent runs.
*
* The rule is "dialog = you are confirming an assignment", NOT "you are
* confirming N runs" (MUL-5010). It therefore does no pre-flight prediction:
* opening it fires no request, so the note box and buttons are usable on the
* first frame. Previously it called POST /api/issues/preview-trigger on open
* and blocked the whole dialog behind a "检查中…" spinner; because that query is
* keyed per issue id with staleTime 0, every new issue was a guaranteed cache
* miss and the wait was unavoidable.
*
* Completion is silent: the assignee change and any run it starts surface
* through the issue's normal assignee / run-status updates, so the confirm adds
* no result toast. Whether a run starts stays the server's existing decision at
* write time. Dismissing the dialog (X / Esc / click-outside) cancels without
* any write. Shared by single assign (1 id) and batch assign (N ids).
*/
export function RunConfirmModal({
onClose,
data,
}: {
onClose: () => void;
data: Record<string, unknown> | null;
}) {
const { t } = useT("modals");
const { getActorName } = useActorName();
const d = (data ?? {}) as RunConfirmData;
const issueIds = d.issueIds ?? [];
const [note, setNote] = useState("");
// Which footer action is in flight, so only the clicked button shows a
// spinner (the request runs an agent on the server for note assigns, so it is
// not instant — the disabled-only state read as frozen).
const [pendingAction, setPendingAction] = useState<"go" | "suppress" | null>(null);
const submitting = pendingAction !== null;
const updateIssue = useUpdateIssue();
const batchUpdate = useBatchUpdateIssues();
// Handoff-support verdict, resolved entirely from warm client caches
// (useWorkspacePresencePrefetch keeps agents / squads / runtimes hot), so the
// note box settles on the first frame with no round-trip — the same shape as
// the quick-create version gate. An agent assignee targets its own runtime; a
// squad targets its leader's, which the squad list gives us directly, so both
// are knowable locally. `null` means "cannot tell" (assignee not in cache
// yet, or no runtime bound) and leaves the box enabled: the note is a soft
// gate, and a spurious warning is worse than a note an old daemon drops.
const wsId = useWorkspaceId();
const { data: agents = [] } = useQuery({ ...agentListOptions(wsId), enabled: !!wsId });
const { data: runtimes = [] } = useQuery({ ...runtimeListOptions(wsId), enabled: !!wsId });
const { data: squads = [] } = useQuery({ ...squadListOptions(wsId), enabled: !!wsId });
const localHandoff = useMemo<boolean | null>(() => {
if (!d.assigneeId) return null;
let agentId: string | undefined;
if (d.assigneeType === "agent") {
agentId = d.assigneeId;
} else if (d.assigneeType === "squad") {
// A squad run is executed by its leader, so the leader's runtime is the
// one that has to render the note.
agentId = squads.find((s) => s.id === d.assigneeId)?.leader_id;
}
if (!agentId) return null;
const agent = agents.find((a) => a.id === agentId);
if (!agent?.runtime_id) return null;
const runtime = runtimes.find((r) => r.id === agent.runtime_id);
if (!runtime) return null;
return handoffSupported(readRuntimeCliVersion(runtime.metadata));
}, [d.assigneeType, d.assigneeId, agents, runtimes, squads]);
// Soft gate: an old runtime can't render the note. Disable the box but let
// the assignment proceed (MUL-3375 §6.3).
const noteDisabled = localHandoff === false;
const applyTo = (extra: Partial<UpdateIssueRequest>) => {
const base: UpdateIssueRequest = {
assignee_type: d.assigneeType ?? null,
assignee_id: d.assigneeId ?? null,
};
return { ...base, ...extra };
};
// The copy names whoever the issue is handed to; for a squad that is the
// squad itself, since its leader deciding who works is an internal detail.
const assigneeName =
d.assigneeName ??
getActorName(d.assigneeType === "squad" ? "squad" : "agent", d.assigneeId ?? "");
const submit = async (suppressRun: boolean) => {
if (issueIds.length === 0 || submitting) return;
setPendingAction(suppressRun ? "suppress" : "go");
const payload = applyTo({
...(suppressRun ? { suppress_run: true } : {}),
...(!suppressRun && !noteDisabled && note.trim() ? { handoff_note: note.trim() } : {}),
});
try {
// Completion is silent, exactly as before: the assignee and any run show
// up through the issue's normal assignee / run-status updates, so there is
// no result toast to add here. Whether a run started is the server's
// existing decision at write time, not something this dialog reports.
if (issueIds.length === 1) {
await updateIssue.mutateAsync({ id: issueIds[0]!, ...payload });
} else {
await batchUpdate.mutateAsync({ ids: issueIds, updates: payload });
}
onClose();
} catch (err) {
toast.error(err instanceof Error && err.message ? err.message : t(($) => $.run_confirm.toast_failed));
setPendingAction(null);
}
};
// States the action, not a prediction: the assignment is certain, the run is
// conditional, so the copy names no run count.
const headline: ReactNode = boldName(
issueIds.length > 1
? t(($) => $.run_confirm.assign_batch, {
name: `${NAME_FENCE}${assigneeName}${NAME_FENCE}`,
count: issueIds.length,
})
: t(($) => $.run_confirm.assign_single, {
name: `${NAME_FENCE}${assigneeName}${NAME_FENCE}`,
}),
);
return (
<Dialog open onOpenChange={(v) => { if (!v && !submitting) onClose(); }}>
<DialogContent>
<DialogHeader>
<DialogTitle>{t(($) => $.run_confirm.title_assign)}</DialogTitle>
<DialogDescription>{headline}</DialogDescription>
</DialogHeader>
{/* Always mounted and always usable on the first frame — nothing about
this box depends on a server answer. */}
<div className="grid gap-1.5">
<label className="text-body font-medium" htmlFor="handoff-note">
{t(($) => $.run_confirm.note_label)}
</label>
<Textarea
id="handoff-note"
value={note}
maxLength={MAX_HANDOFF_NOTE}
disabled={submitting || noteDisabled}
placeholder={t(($) => $.run_confirm.note_placeholder)}
onChange={(e) => setNote(e.target.value)}
rows={3}
/>
{noteDisabled ? (
<p className="text-caption text-muted-foreground">{t(($) => $.run_confirm.note_unsupported)}</p>
) : null}
</div>
{/* The only spinner left is on the button the user just pressed, and it
reflects the write in flight — never a pre-flight check. */}
<DialogFooter>
<Button type="button" variant="outline" disabled={submitting} onClick={() => submit(true)}>
{pendingAction === "suppress" ? <Spinner className="size-4" /> : t(($) => $.run_confirm.dont_start)}
</Button>
<Button type="button" disabled={submitting} onClick={() => submit(false)}>
{pendingAction === "go" ? <Spinner className="size-4" /> : t(($) => $.run_confirm.confirm_assign)}
</Button>
</DialogFooter>
</DialogContent>
</Dialog>
);
}