ui: round three — the tight band, one geometry everywhere

Owner round three on the user rows: (1) the band padding halves,
py-2.5 -> py-[5px] — subtle, not tall; same for the messages. (2) The
in-card thinking joins the band: identical styling and heights for
collapsed thinking in the chat and inside question cards (card-on-card
is quiet by design; revert if it reads too light). (3) The left gap
MATCHES the new 5px padding, so the rail fills the rest of the 3ch
inset - border-l-[calc(3ch-5px)] pl-[5px]: a thick muted rail, total
inset still 3ch (the chevron-alignment law survives). The banded prop
dies - one geometry, cards included.
This commit is contained in:
amosblomqvist
2026-10-08 18:32:21 +03:00
parent 3691e07118
commit 497b29804c
6 changed files with 39 additions and 51 deletions
+5 -10
View File
@@ -143,16 +143,11 @@ export function AnsweredCard({ answer }: { answer: QuestionAnswerData }) {
{(answer.thinking ?? answer.note) && (
/* The SAME unified row (owner round, 2026-10-05): the quiz answer's
thinking is the identical collapsible ThinkingBlock — it just
happens to sit INSIDE the card, so it takes the rail but no band
(owner round two, 2026-10-07: card-color on card-color would be
invisible; geometry alone carries it there). */
<ThinkingBlock
testId="your-thinking"
label="YOUR THINKING"
text={answer.thinking ?? answer.note ?? ""}
banded={false}
/>
thinking is the identical collapsible ThinkingBlock — and since
owner round three (2026-10-07) IDENTICALLY styled, band and all:
collapsed thinking matches in height everywhere, card or chat
(card-on-card is quiet by design; too light → revert). */
<ThinkingBlock testId="your-thinking" label="YOUR THINKING" text={answer.thinking ?? answer.note ?? ""} />
)}
{/* px-3→removed: card p-4 + box px-3 double-inset sat the "Why" label deeper
+6 -5
View File
@@ -1399,13 +1399,14 @@ export function Chat({
Rendered WITHOUT the Message wrapper — that is the bubble. */
<div
data-testid="user-message"
/* THE DOCUMENT, ALL-IN (owner round two, 2026-10-07): no
chat side — the user message is a full-width card-color
BAND with the 6px rail, the same geometry as the thinking
rows (the band is the authorship mark against the agent's
/* THE DOCUMENT, ALL-IN (owner rounds two+three, 2026-10-07):
no chat side — the user message is a full-width card-color
BAND, the IDENTICAL geometry as the thinking rows: 5px band
padding, the left gap matches it, the rail fills the 3ch
inset (the band is the authorship mark against the agent's
transparent prose). The "You" label stays — one register
with the thinking labels. */
className="my-2 w-full border-l-[6px] border-muted-foreground/30 bg-card py-2.5 pl-[calc(3ch-6px)] pr-2 font-ui text-xs"
className="my-2 w-full border-l-[calc(3ch-5px)] border-muted-foreground/30 bg-card py-[5px] pl-[5px] pr-2 font-ui text-xs"
>
{/* The rail GROUPS (owner round 4, 2026-10-05): the YOU heading
sits INSIDE the rail, same mono-xs register and same 1ch
+18 -20
View File
@@ -17,17 +17,19 @@ import { cn } from "@/lib/utils";
* The glance-level law (owner): user thinking = rail + 50% italic;
* agent thinking = no rail.
*
* THE DOCUMENT, ALL-IN (owner round two, 2026-10-07): no chat sides —
* transcript-level user rows (this block, the "You" message) are
* full-width CARD-COLOR BANDS: the band is the authorship mark against
* the agent's transparent prose, and the rail runs along the band's left
* edge. The rail keeps its muted color but is now 6px wide, eating the
* artificial air of the old 3ch inset — the total inset STAYS 3ch so the
* chevron-alignment law with the agent's flush rows survives; the gap
* between rail and text drops to natural (3ch - 6px). In-card rows
* (inside the answered card, on the card's own background) take the rail
* but NO band (`banded: false`). `defaultOpen` marks transcript rows,
* which arrive EXPANDED (clearly conveyed that the thinking was sent).
* THE DOCUMENT, ALL-IN (owner rounds two+three, 2026-10-07): no chat
* sides — user rows are full-width CARD-COLOR BANDS (the band is the
* authorship mark against the agent's transparent prose) with the rail
* along the left edge. ONE geometry everywhere, in cards and out: the
* band padding is a tight 5px (round three: halved — subtle, not tall),
* and the left gap MATCHES it, so the rail fills the rest of the 3ch
* inset — border-l-[calc(3ch-5px)] pl-[5px]. Total inset stays 3ch (the
* chevron-alignment law with the agent's flush rows survives); the rail
* grows as the padding shrinks. In-card rows carry the identical
* styling (owner round three: same band, same heights — card-on-card is
* quiet by design; if it reads too light we revert). `defaultOpen`
* marks transcript rows, which arrive EXPANDED (clearly conveyed that
* the thinking was sent).
*
* ⌘T (toggle-user-thinking) sets EVERY row to one state via the module
* store below — one source of truth, no provider tree: rows subscribe on
@@ -66,7 +68,6 @@ export function ThinkingBlock({
text,
testId,
dataCommit,
banded = true,
defaultOpen = false,
}: {
label?: string;
@@ -74,8 +75,6 @@ export function ThinkingBlock({
testId?: string;
/** The practice marker's register flag (commit vs nudge). */
dataCommit?: boolean;
/** Transcript rows carry the card-color band; in-card rows rail only. */
banded?: boolean;
/** Transcript rows arrive expanded (owner, 2026-10-07). */
defaultOpen?: boolean;
}) {
@@ -101,13 +100,12 @@ export function ThinkingBlock({
data-testid={testId}
data-commit={dataCommit === undefined ? undefined : dataCommit ? "true" : "false"}
className={cn(
// The rail geometry (owner round two, 2026-10-07): total inset
// The rail geometry (owner round three, 2026-10-07): total inset
// STAYS 3ch (the chevron-alignment law with the agent's flush
// rows), re-split — 6px rail (was 2px) + a natural 3ch-6px gap
// (was the too-wide 3ch-2px). In-card rows keep the 12px my-3
// rhythm; banded rows sit my-2 apart (the band's own air).
"lh-card-enter w-full border-l-[6px] border-muted-foreground/30 pl-[calc(3ch-6px)] font-ui text-xs",
banded ? "my-2 bg-card py-2.5" : "my-3",
// rows). The band padding is a tight 5px (halved from 10px) and
// the left gap MATCHES it — the rail fills the rest:
// calc(3ch-5px) wide. One geometry everywhere, cards included.
"lh-card-enter my-2 w-full border-l-[calc(3ch-5px)] border-muted-foreground/30 bg-card py-[5px] pl-[5px] font-ui text-xs",
)}
>
<button
+5 -4
View File
@@ -295,13 +295,14 @@ describe("render smoke tests", () => {
// trigger row only — the thinking text is one click away, never loud.
assert.ok(html.includes('data-testid="your-thinking-trigger"'), "the trigger row");
assert.ok(html.includes('aria-expanded="false"'), "arrives collapsed");
// IN-CARD = RAIL ONLY (owner round two, 2026-10-07): no band on the
// card's own background; geometry alone carries it there.
// IDENTICAL GEOMETRY (owner round three, 2026-10-07): the in-card row
// carries the band too — collapsed thinking matches in height
// everywhere, card or chat.
const yourRow = html.slice(html.indexOf('data-testid="your-thinking"'));
assert.ok(!yourRow.slice(0, 400).includes("bg-card"), "no band inside the card");
assert.ok(yourRow.slice(0, 400).includes("bg-card"), "the band inside the card as well");
// THE RAIL GROUPS (owner correction): the label sits INSIDE the rail —
// ownership readable from the rail alone, collapsed or expanded.
const railAt = html.indexOf("border-l-[6px]");
const railAt = html.indexOf("border-l-[calc(3ch-5px)]");
const trigAt = html.indexOf("your-thinking-trigger");
assert.ok(railAt !== -1 && railAt < trigAt, "the rail opens before the label (same quote)");
assert.ok(!html.includes("brightness isn&#x27;t distance"), "the text is hidden until expanded");
+4 -11
View File
@@ -242,24 +242,17 @@ describe("md-blocks — the restored plain blockquote (owner round, 2026-10-07)"
});
});
describe("ThinkingBlock — side-based authorship + ⌘T sync (owner round, 2026-10-07)", () => {
it("the banded row is the document band — full-width card color, 6px rail", () => {
describe("ThinkingBlock — the document band + ⌘T sync (owner rounds, 2026-10-07)", () => {
it("the row is the document band — one geometry, cards included (owner round three)", () => {
const html = renderToStaticMarkup(<ThinkingBlock text="banded" defaultOpen />);
assert.ok(html.includes("bg-card"), "the card-color band");
assert.ok(html.includes("border-l-[6px]"), "the 6px rail");
assert.ok(html.includes("pl-[calc(3ch-6px)]"), "natural gap, 3ch total inset preserved");
assert.ok(html.includes("border-l-[calc(3ch-5px)]"), "the rail fills the 3ch inset");
assert.ok(html.includes("py-[5px]") && html.includes("pl-[5px]"), "tight 5px padding, gap matches it");
assert.ok(html.includes("w-full"), "full width — no chat side");
assert.match(html, /aria-expanded="true"/);
assert.ok(html.includes("banded"), "text visible on arrival");
});
it("in-card rows take the rail but no band", () => {
const html = renderToStaticMarkup(<ThinkingBlock text="in card" banded={false} />);
assert.ok(!html.includes("bg-card"), "no band on the card's own background");
assert.ok(html.includes("border-l-[6px]"), "the rail stays");
assert.ok(html.includes("my-3"), "the 12px in-card rhythm");
});
it("⌘T sets every row — explicit sets stomp individual state, late mounts honor the last set", () => {
setAllUserThinkingOpen(false);
const collapsed = renderToStaticMarkup(<ThinkingBlock text="synced" defaultOpen />);
+1 -1
View File
@@ -185,7 +185,7 @@ describe("the practice transcript cards", () => {
assert.match(trace, /aria-expanded="true"/);
assert.ok(trace.includes("stuck at step two"), "the text is visible on arrival");
assert.ok(trace.includes("bg-card"), "the card-color band");
assert.ok(trace.includes("border-l-[6px]"), "the 6px rail, left side");
assert.ok(trace.includes("border-l-[calc(3ch-5px)]"), "the rail fills the 3ch inset");
const legacy = renderToStaticMarkup(<ThinkingCard data={{ problemCallId: "c", commit: true, text: "It is 42." }} />);
assert.match(legacy, /YOUR THINKING/);
assert.doesNotMatch(legacy, /Your solution/);