/**
 * OTA content page mast.
 *
 * The Hero component set from the OTA design file, node 241:7394, as the band
 * that replaces the Agend theme's own mast. Four post types share it: content
 * pages and interest groups against a square media slot, news and events
 * against a wider one. Only the content page and its empty states are built
 * here; see the note on the media slot for what the other two change.
 *
 * EVERY COLOUR IS A ROLE, AND EVERY ROLE ALREADY EXISTED
 *
 * The frame binds the whole band, so unlike the search page there is no reading
 * by hex to do here and no orphan greys to argue about. Binding for binding:
 *
 *   band background    bar/background     --ota-background
 *   breadcrumb link    bar/accent         --ota-accent
 *   breadcrumb rest    bar/text           --ota-text
 *   eyebrow            bar/accent         --ota-accent
 *   title              bar/White Text     --ota-white-text
 *   intro              bar/text           --ota-text
 *   brand mark         bar/overlay        --ota-overlay
 *
 * The frame binds that last one to a solid ochre staircase filling the media
 * slot. The staircase is gone: too heavy against real copy, replaced on
 * 14 Aug 2026 with the OTA brand mark carrying the footer's own treatment. Same
 * role, so it still follows the theme; see the mark rule at the bottom.
 *
 * `--ota-white-text` for the title is not a mistake and not a duplicate of
 * `--ota-text`. The two diverge: Teal resolves the title to #FFFFFF and the
 * intro to #F0E7D8, Ash resolves both to #003038. Verified against the hero
 * instances in the Teal and Ash page frames, 241:7383 and 268:7441.
 *
 * ONE HEIGHT, AND IT IS A FLOOR RATHER THAN A HEIGHT
 *
 * The frame draws 464, 300 without the intro, 496 on news and 559 on an event.
 * None of those is a number anybody chose, confirmed with Bryce on 14 Aug 2026:
 * the band should grow with its content. So `--oph-media` is a minimum, not a
 * size. The body hugs its own content, the band takes the larger of that and
 * the floor, and the media then stretches to whatever the band ended up being.
 * Long titles and long intros push it past 464; short ones cannot pull it under.
 *
 * Measured against the real thing, the frame's own copy makes a body 312px
 * tall, so the 464 it draws is this floor rather than the text. That is worth
 * knowing before anyone "fixes" this to hug more tightly.
 *
 * THE MEDIA STRETCHES, IT DOES NOT SET THE HEIGHT. An earlier pass had the
 * media as a fixed square that stood the band up, which meant a tall band left
 * it floating with a strip of background above and below. The floor does that
 * job now, so the image fills the band edge to edge at any height.
 *
 * WHERE THIS DELIBERATELY LEAVES THE FRAME: a page with no intro. The frame
 * draws that at 300 and crops the image to a letterbox to get there. We hold
 * 464. Shortening the band is the easy half; the hard half is that these images
 * are transparent PNGs cut to a silhouette, so a shorter band trims the
 * silhouette rather than showing less of a photograph. If the 300 is wanted
 * back it has to come with an answer for that.
 *
 * THE MEASURES ARE REAL CONSTRAINTS, NOT ARTBOARD LEFTOVERS. The frame caps the
 * title at 810px and the intro at 618px inside a body column that is 896px wide
 * at 1440, so neither is the column simply running out of room. Both are line
 * length decisions rather than incidental widths.
 *
 * The title keeps the frame's 810. The intro does not keep the frame's 618: it
 * was widened to 810 on 10 Sep 2026 so the two share an edge, because the 618
 * cap is fixed while the column grows and the dead space to its right grew with
 * the viewport. The full reasoning sits on the `.ota-page-hero__intro` rule.
 *
 * WHAT NEWS AND EVENTS WILL CHANGE, when they are built: `--oph-media` goes to
 * 499px and the media takes `border-start-start-radius` and
 * `border-end-start-radius` of `--radius/md`, 14px, on its left corners only.
 * News also top aligns its column instead of centring it, which is the one
 * genuine structural difference between the four. Neither is stubbed here,
 * because dead rules for a component nobody has built yet rot faster than they
 * save time.
 *
 * FONTS ARE INHERITED, NOT DECLARED. The band is NB International Pro
 * throughout in the frame, which is already the site's body face, so naming it
 * again here would be a second place to change it. Only weight, size and
 * tracking are set. The intro's 18/28.8 Bold matches no registered type token:
 * `Body/Lead` is 22 Regular and `Body/Default` is 17 Light, so it is held
 * literal rather than bent onto a token that is not it.
 */

.ota-page-hero {
	--oph-surface: var(--ota-background, #f0e7d8);
	--oph-text: var(--ota-text, #003038);
	--oph-heading: var(--ota-white-text, #003038);
	--oph-accent: var(--ota-accent, #f04c3a);
	--oph-mark: var(--ota-overlay, #f04c3a);

	/*
	 * The media column's width, and the band's minimum height. One number doing
	 * both jobs because the frame's slot is square. News and events widen the
	 * column to 499px, at which point the two part company and this needs
	 * splitting in two.
	 */
	--oph-media: 464px;

	/* `radius/md` in the frame. Left corners only, see the media rule. */
	--oph-radius: 14px;

	/*
	 * The site's own container, and the design's inset inside it.
	 *
	 * The band bleeds the full width of the viewport, but its copy must not:
	 * measured from the viewport edge, an 80px gutter sits under the logo at
	 * 1440 and then walks away from every other element on the page as the
	 * window widens, which is what it was doing.
	 *
	 * 1440 and not 1140. The theme runs two containers: the header centres on
	 * 1440 with 32px of padding, and Elementor's own sections centre on 1140.
	 * Derived by measuring the logo at five widths, where its left edge is
	 * always `(100vw - 1440) / 2 + 32`. The band is chrome, it sits against the
	 * header, and the frame draws it flush with the header rather than with the
	 * body copy, so it follows the header.
	 *
	 * The 48px offset between this gutter and the logo's 32 is in the design
	 * too, so it is kept rather than flattened: the frame puts the logo at 32
	 * and the crumbs at 80 on the same 1440 artboard.
	 */
	--oph-max: 1440px;
	--oph-gutter: 80px;

	/*
	 * 92 and not 48. The 48 came from the `pb-[var(--spacing/48,48px)]` in the
	 * frame's generated code, which is padding on an inner block rather than the
	 * band's own, and taking it literally made every band 88px shorter than
	 * drawn. The real figure is in the geometry: the `Text=No` variant puts its
	 * text block at y=91 with 118 of content in a 300 band, so 91 + 118 + 91,
	 * and the `Text=Yes` variant does the same sum at 95.5 with a taller block.
	 *
	 * This is what makes the design's own heights fall out for free rather than
	 * being propped up. 114 of content gives 298, which is the 300 the frame
	 * draws for a page with no intro.
	 */
	--oph-pad: 92px;

	/* The mark is positioned against this. */
	position: relative;
	display: flex;
	align-items: stretch;
	justify-content: space-between;
	min-block-size: var(--oph-media);
	background-color: var(--oph-surface);

	/*
	 * `clip` and not `hidden`: the band never scrolls, and `hidden` on a flex
	 * parent creates a scroll container that swallows any focus ring on the
	 * crumbs sitting near its edge.
	 */
	overflow: clip;

	/*
	 * TWO PIXELS OF ITS OWN COLOUR ABOVE ITSELF, WHICH IS NOT DECORATION.
	 *
	 * The site header is `position: fixed` and this band butts directly against
	 * it, and `html`, `body` and `#agendtheme-content` are all white while the
	 * band paints its own theme. So anything that opens even a hairline between
	 * the two shows WHITE, which is the flash Bryce reported on 17 Aug 2026
	 * when scrolling back up.
	 *
	 * A fixed element and the scrolled content beneath it are composited on
	 * separate layers, and at fractional scroll offsets Chrome can round their
	 * edges to different device pixels. This band's own height is fractional
	 * too, 379.047 and 297.453 on the routes measured, which is the ideal
	 * ingredient for that.
	 *
	 * BE HONEST ABOUT THIS ONE: it is a mitigation, not a confirmed fix. Logged
	 * out, at eight device scale factors and ten scroll offsets, no seam
	 * appears, so this could not be reproduced to prove the cure. It is applied
	 * because the cost is nil rather than because the cause is nailed down.
	 *
	 * Nil because the header reads the same `--ota-background` role this band
	 * does, so the 2px is the colour it is already sitting on. It cannot show
	 * as a line even if the seam never existed.
	 */
	box-shadow: 0 -2px 0 0 var(--oph-surface);
}

/*
 * THE FLOOR IS FOR THE IMAGE, NOT FOR THE BAND, so a page without one does not
 * pay for it.
 *
 * `--oph-media` is a minimum height only because the media column stretches to
 * whatever the band turns out to be, and a feature image is a transparent PNG
 * cut to a silhouette: let the band get short and the column goes letterbox and
 * slices the top and bottom off the shape. None of that applies here. With no
 * image there is no column, the brand mark is positioned against the band and
 * scales with it, and the floor was doing nothing but padding an empty band out
 * to 464 when its content wanted 298.
 *
 * So this is the state where the frame's own numbers and a content driven band
 * finally agree: 300 drawn, 298 built.
 */
.ota-page-hero--no-media {
	min-block-size: 0;
}

/* -------------------------------------------------------------------------
 * The four routes.
 * ---------------------------------------------------------------------- */

/*
 * NEWS AND EVENTS: ONE NUMBER STILL DOES BOTH JOBS, which is not what the note
 * at the top of this file predicted. It expected the column width and the band's
 * floor to part company at 499px and need splitting in two. Measured, they do
 * not: the frame draws the news band 499 wide and 496 tall, near enough square,
 * and draws the event band taller than 499 only because its copy is taller. So
 * 499 is both the column and the floor here exactly as 464 is on a page, and the
 * event's 559 falls out of content the way every other height in this file does.
 *
 * The floor still earns its place for a different reason than on a page. These
 * are photographs cropped with `cover`, not silhouettes, so a short band does
 * not slice a shape in half; it just turns a 499px wide image into a letterbox
 * strip. The floor is what stops that on a post with a one line title and no
 * excerpt.
 */
.ota-page-hero--news,
.ota-page-hero--event,
.ota-page-hero--events {
	--oph-media: 499px;
}

/*
 * THE VERTICAL PADDING IS PER ROUTE, AND THE THREE NUMBERS ARE MEASURED RATHER
 * THAN CHOSEN. Same arithmetic as the page's 92: read the frame's content block
 * out of its band and halve what is left.
 *
 *   page   300 band, 118 of content   ->  91
 *   news   496 band, 401 of content   ->  47.5
 *   event  559 band, 406 of content   ->  76.5
 *
 * They differ because the bands were drawn at different times around different
 * copy, not because anyone specified three paddings. Kept as measured anyway:
 * flattening them to one number moves all three bands off the frame, and there
 * is no evidence for which of the three would be the right survivor.
 */
.ota-page-hero--news {
	--oph-pad: 48px;
}

.ota-page-hero--event {
	--oph-pad: 76px;
}

.ota-page-hero__body {
	/* Above the mark, which reaches under the copy on a narrow viewport. */
	position: relative;
	z-index: 1;
	display: flex;
	flex: 1 1 0;
	flex-direction: column;
	gap: 24px;
	align-items: flex-start;
	justify-content: center;
	min-inline-size: 0;
	padding-block: var(--oph-pad);

	/*
	 * Half of whatever the viewport has over the container, plus the design's
	 * gutter. Below 1440 the `max()` floors the first term at zero and this is
	 * just the gutter, so nothing changes at or under the drawn width.
	 *
	 * The media is deliberately NOT given the matching treatment on the right.
	 * It keeps running to the edge of the viewport, the way the home page hero
	 * slider's artwork does, because stopping it at the container line would
	 * leave a strip of background outside it that reads as a mistake.
	 */
	padding-inline: calc(max(0px, (100% - var(--oph-max)) / 2) + var(--oph-gutter)) 40px;
}

.ota-page-hero__crumbs {
	display: flex;
	flex-wrap: wrap;
	align-items: center;
	gap: 10px;

	/*
	 * `spacing/8` sits on the crumb row in the frame, which is what makes that
	 * block 30 tall rather than 22. It is load bearing now that the band's
	 * height is the sum of its content: without it every band is 8px under.
	 */
	padding-block-start: 8px;
	font-size: 13px;
	font-weight: 700;
	line-height: 21.45px;
	color: var(--oph-text);
}

.ota-page-hero__crumb {
	color: inherit;
	text-decoration: none;
}

/*
 * ONLY THE FIRST CRUMB IS ACCENT, and it is the first rather than every link.
 * The frame paints Home in `bar/accent` and every crumb after it in `bar/text`,
 * including the intermediate ones that are just as clickable. So this is a
 * position, not a state: `a` here would light up Practice Support too and the
 * trail would read as two colours of link.
 */
.ota-page-hero__crumb:first-child {
	color: var(--oph-accent);
}

/*
 * The colour is declared here and not only at rest, because the theme's bare
 * `a:hover` (0,1,1) outguns `.ota-page-hero__crumb` (0,1,0) in the hover state
 * and repainted the trail in the page's link-hover colour, which on a dark
 * band is the band's own navy: the link vanished under the pointer. The frame
 * draws hover as the underline alone, colours unchanged.
 */
.ota-page-hero__crumbs > a.ota-page-hero__crumb:hover,
.ota-page-hero__crumbs > a.ota-page-hero__crumb:focus-visible {
	color: var(--oph-text);
	text-decoration: underline;
}

/* Home keeps its accent under the pointer, same position-not-state reasoning as above. */
.ota-page-hero__crumbs > a.ota-page-hero__crumb:first-child:hover,
.ota-page-hero__crumbs > a.ota-page-hero__crumb:first-child:focus-visible {
	color: var(--oph-accent);
}

/*
 * The separator is drawn rather than printed into the markup. `+` rather than
 * `:not(:first-child)` so it survives the trail being built with the current
 * page omitted.
 *
 * THE OLD NOTE HERE WAS WRONG AND IS WORTH CORRECTING RATHER THAN DELETING. It
 * claimed CSS generated content keeps the slash away from a screen reader. It
 * does not: every current browser exposes `content` in the accessibility tree,
 * so VoiceOver and NVDA were announcing "slash" between every pair of crumbs.
 * Caught by Nathan in review on PR #5.
 *
 * The two value syntax is the fix, `content: "/" / ""`, where the part after the
 * slash is the accessible name. An empty string there means the glyph is drawn
 * and nothing is announced. The plain declaration stays directly above it as the
 * fallback for browsers that do not parse the alt text form, which drop the
 * whole declaration rather than half of it.
 */
.ota-page-hero__crumb + .ota-page-hero__crumb::before {
	content: "/";
	content: "/" / "";
	margin-inline-end: 10px;
	color: var(--oph-text);
}

.ota-page-hero__heading {
	display: flex;
	flex-direction: column;
	gap: 8px;
	align-items: flex-start;
}

.ota-page-hero__eyebrow {
	margin: 0;
	font-size: 17px;
	font-weight: 700;
	line-height: 24.75px;
	letter-spacing: 1px;
	color: var(--oph-accent);
}

/*
 * Fluid rather than the frame's flat 60px. The frame only draws 1440 and 393,
 * and 60px on a 393 viewport puts a two word title on three lines.
 *
 * BOTH ENDS ARE REGISTERED TYPE TOKENS, WHICH IS WHY THEY ARE THE ANCHORS AND
 * NOT ROUND NUMBERS: `Heading/H1` is 3.75rem and `Heading/H3` is 2rem in the
 * kit, so the clamp runs between two sizes the site already has rather than
 * inventing a range. Checked against the kit, not the frame.
 *
 * The slope is solved for those two points and not eyeballed: 28px of growth
 * over 1047px of viewport is 2.674vw, and 60 - (0.02674 x 1440) leaves the
 * 1.343rem. So it lands on exactly 32px at 393 and exactly 60px at 1440, where
 * the max caps it, which is also what holds the title at the frame's size on a
 * monitor wider than the artboard instead of letting it keep growing.
 *
 * An earlier pass used `2.2vw + 1.44rem`, which put 54.7px at 1440: the max was
 * doing the work of the anchor and the title only reached the frame's 60px at
 * 1680. It read as nearly right at every width, which is how it survived. If
 * this is ever retuned, check the value AT 1440 rather than trusting the cap.
 */
.ota-page-hero__title {
	max-inline-size: 810px;
	margin: 0;
	font-size: clamp(2rem, 2.674vw + 1.343rem, 3.75rem);
	font-weight: 700;
	line-height: 1;
	letter-spacing: -0.01em;
	color: var(--oph-heading);
}

/*
 * MATCHED TO THE TITLE'S CAP, NOT THE FRAME'S 618.
 *
 * The frame caps the intro at 618 inside an 896 body column at 1440, which is
 * a sound line length on its own (about 73 characters). The problem is that
 * the cap is fixed while the body column grows with the viewport, so the dead
 * space to the right of the intro grows with it: 78px unused at 1280, 238 at
 * 1440, 478 at 1920, 798 at 2560. On a 1280 laptop it looks correct, which is
 * why it went unnoticed.
 *
 * Jade, 10 Sep 2026 (PA 49497888): the subheader "should sit a little wider
 * across the banner... the text and image is sitting a little too far apart".
 * Her red mark measures to a text width of about 724px at her viewport.
 *
 * 810 rather than 724 because it is the title's existing cap directly above,
 * so the two now share an edge instead of ending at two arbitrary widths, and
 * it clears her mark rather than landing just short of it. Still a cap, not
 * fluid: at 2560 there is dead space again, deliberately, because a line
 * running the full column would be far past a readable length.
 */
.ota-page-hero__intro {
	max-inline-size: 810px;
	margin: 0;
	font-size: 18px;
	font-weight: 700;
	line-height: 28.8px;
	color: var(--oph-text);
}

/* -------------------------------------------------------------------------
 * Pills: the event's date, node 451:23150, and news Share and Print, 2587:34034.
 * ---------------------------------------------------------------------- */

/*
 * One shape, two weights. The frame draws the date at a 1.5px border and the two
 * news controls at 1px, which is the only thing separating them, so the border
 * is the variable and everything else is shared.
 *
 * `inline-flex` and `align-self: flex-start` because the body is a flex column
 * with `align-items: flex-start`: without it a pill stretches the full width of
 * the column and its 999px radius turns into a stadium the width of the band.
 */
.ota-page-hero__pill {
	--oph-pill-border: 1.5px;

	display: inline-flex;
	gap: 8px;
	align-items: center;
	align-self: flex-start;
	margin: 0;
	border: var(--oph-pill-border) solid var(--oph-text);
	border-radius: 999px;
	padding: 10px 20px;
	background: none;
	font: inherit;
	font-size: 14.4px;
	font-weight: 500;
	line-height: 23.76px;
	color: var(--oph-text);
}

/*
 * Share and Print. A `<button>`, so this has to say `font: inherit` above and
 * undo the user agent's own idea of a control; `cursor` because a button that
 * does something in the page still reads as clickable.
 */
.ota-page-hero__pill--action {
	--oph-pill-border: 1px;

	padding: 8px 16px;
	font-size: 14px;
	line-height: 23.1px;
	cursor: pointer;
}

/*
 * The mark inside a pill. Masked rather than an `<img>` for the reason the brand
 * mark gives: `currentColor` then carries the theme's text role into it, so the
 * calendar on a Teal event is sand and on an Ash one is navy without a second
 * file. The three source files bake their own stroke colour, which the mask
 * discards along with everything except the alpha.
 */
.ota-page-hero__pill-icon {
	flex: 0 0 auto;
	inline-size: 15px;
	block-size: 15px;
	background-color: currentcolor;
	-webkit-mask: var(--oph-icon) no-repeat center / contain;
	mask: var(--oph-icon) no-repeat center / contain;
}

.ota-page-hero__pill-icon--calendar {
	--oph-icon: url("../img/hero/calendar.svg");
}

/* The two news marks are drawn at 14 rather than the calendar's 15. */
.ota-page-hero__pill-icon--share,
.ota-page-hero__pill-icon--print {
	inline-size: 14px;
	block-size: 14px;
}

.ota-page-hero__pill-icon--share {
	--oph-icon: url("../img/hero/share.svg");
}

.ota-page-hero__pill-icon--print {
	--oph-icon: url("../img/hero/print.svg");
}

/* -------------------------------------------------------------------------
 * The news byline row. Node 2587:34022.
 * ---------------------------------------------------------------------- */

.ota-page-hero__meta {
	display: flex;
	flex-wrap: wrap;
	gap: 16px;
	align-items: center;
	font-size: 14px;
	font-weight: 300;
	line-height: 23.1px;
	color: var(--oph-text);
}

.ota-page-hero__byline {
	display: inline-flex;
	gap: 10px;
	align-items: center;
}

/* The name is the only bold thing in the row; "By" stays Light. */
.ota-page-hero__byline b {
	font-weight: 700;
}

/*
 * 34px and round. `get_avatar()` is asked for 68 so the file is right on a 2x
 * screen, and sized down here rather than in the request, which is what gives
 * the browser a sharp one either way.
 */
.ota-page-hero__avatar {
	flex: 0 0 auto;
	inline-size: 34px;
	block-size: 34px;
	border-radius: 999px;
	object-fit: cover;
}

/*
 * THE TWO HAIRLINES, DRAWN RATHER THAN PRINTED. The frame puts a 1px rule
 * between the byline and the date, and another between the date and the
 * controls. They are `::before` on what follows them so there is no empty
 * element in the row for a screen reader to walk past, the same reasoning the
 * crumb separator uses further up.
 *
 * `--ota-divider` is a real role and every theme sets it, at 20% of the band's
 * own text colour, so this follows the theme rather than needing its own value.
 *
 * The margin is negative on purpose: the row's `gap` is 16 and the frame puts
 * the rule 16 clear on both sides, so the hairline has to sit inside the gap it
 * is dividing rather than adding a third one.
 */
.ota-page-hero__published::before,
.ota-page-hero__actions::before {
	content: "";
	display: inline-block;
	inline-size: 1px;
	block-size: 16px;
	margin-inline-end: 16px;
	background-color: var(--ota-divider, rgba(0, 48, 56, 0.2));
	vertical-align: middle;
}

/*
 * The controls, written in by the script in the template. `gap` matches the
 * row's own so the pills sit the same distance apart as everything else in it.
 */
.ota-page-hero__actions {
	display: inline-flex;
	flex-wrap: wrap;
	gap: 16px;
	align-items: center;
}

/*
 * The row is chrome, not content, so it goes when the page does. The band is the
 * one part of a printed article that has no job: the reader is holding the
 * thing, and Share and Print in particular are controls on paper.
 */
@media print {
	.ota-page-hero__actions {
		display: none;
	}

	/*
	 * THE BAND HAS TO BE RESET FOR INK, OR A TEAL NEWS POST PRINTS BLANK.
	 *
	 * Chrome does not print background colours unless the reader ticks
	 * Background graphics, but it does keep every text colour. On a Teal band
	 * that means the `<h1>` prints in `--ota-white-text`, which this file's own
	 * header note records as #FFFFFF on Teal, onto white paper. The crumbs, the
	 * intro and the byline print in `--ota-text`, sand #F0E7D8, which is barely
	 * darker. A white heading and near-invisible body copy, on a page the Print
	 * pill in this very band invites people to produce.
	 *
	 * Found by Nathan in review, and measured: on `/event/ai-use-in-paediatrics`
	 * under print emulation the title computed to rgb(255, 255, 255).
	 *
	 * So the band drops its own colours and takes ink instead. The mark goes too:
	 * it is a masked background, which prints as nothing anyway, so it can only
	 * ever be a blank gap holding the layout open.
	 */
	.ota-page-hero {
		background: none;
		box-shadow: none;
	}

	/*
	 * `:first-child` IS NAMED EXPLICITLY, because the first crumb has its own
	 * accent colour rule at (0,2,0) further up and a bare `.ota-page-hero__crumb`
	 * here loses to it. Measured after the first pass at this print block: the
	 * title and the pills came out navy as intended and the Home crumb was still
	 * printing ochre #F68B1F.
	 */
	/*
	 * The separator is named at the same `+` compound as its screen rule, for
	 * the reason `:first-child` is: the bare `::before` here is (0,1,1) and the
	 * screen rule's explicit colour at (0,2,1) beat it, so the slashes kept
	 * printing sand. Same trap, one row down. Raised in review on PR #5.
	 */
	.ota-page-hero__title,
	.ota-page-hero__eyebrow,
	.ota-page-hero__intro,
	.ota-page-hero__crumb,
	.ota-page-hero__crumb:first-child,
	.ota-page-hero__crumb::before,
	.ota-page-hero__crumb + .ota-page-hero__crumb::before,
	.ota-page-hero__meta,
	.ota-page-hero__pill {
		color: #003038;
	}

	.ota-page-hero__pill {
		border-color: #003038;
	}

	/*
	 * The pill icon is the same masked background as the mark, so it prints as
	 * nothing too. The date pill sits outside `__actions`, so the hide above
	 * does not catch it, and without this the calendar leaves a blank gap next
	 * to text that now prints correctly. Raised in review on PR #5.
	 *
	 * The byline divider joins them for the same reason: it is a 1px
	 * `background-color` box, dropped by Background-graphics-off printing, so
	 * without this it was an invisible spacer holding 33px open between the
	 * byline and the date. Hidden rather than redrawn as a border: it is
	 * chrome, and the row's own gap keeps the two apart on paper. The
	 * `__actions` divider goes down with the actions hide above. Raised in
	 * review on PR #5.
	 */
	.ota-page-hero__mark,
	.ota-page-hero__pill-icon,
	.ota-page-hero__published::before {
		display: none;
	}
}

/*
 * The right hand column. It takes the band's full height rather than setting
 * it, so the artwork meets the top and bottom edges however tall the copy makes
 * the band.
 */
.ota-page-hero__media {
	flex: 0 0 var(--oph-media);
	align-self: stretch;
	inline-size: var(--oph-media);
}

/*
 * News and events round the photo's leading corners, radius/md 14, node
 * 2587:34077: the image runs off the right edge of the viewport, so only the
 * two corners facing the copy are drawn. The radius clips the COLUMN rather
 * than the img on purpose: the theme forces its own border-radius onto every
 * img under main#agendtheme-content (_content.scss:264, at id strength), and
 * clipping the container wins that fight without entering it.
 */
.ota-page-hero--news .ota-page-hero__media,
.ota-page-hero--event .ota-page-hero__media {
	overflow: hidden;
	border-start-start-radius: 14px;
	border-end-start-radius: 14px;
}

/*
 * TWO CLASSES, FOR SPECIFICITY RATHER THAN STYLE. Elementor ships
 * `.elementor img { height: auto; max-width: 100% }` at (0,1,1), which ties a
 * lone `.ota-page-hero__image` and wins on load order. `height: auto` collapses
 * the box back to the file's intrinsic size, which is the letterboxing this
 * rule exists to stop. Same trap the hero carousel and the category feature
 * both hit; `max-width: none` is here for the same reason.
 *
 * `cover`, following the home page hero slider: fill the column and trim,
 * rather than shrink to fit and leave a band of background above and below.
 *
 * `object-position: center` rather than the carousel's `left center`. That
 * value is tuned to a box that is half the viewport and changes shape with it;
 * this column is a fixed 464 and centring is the neutral answer. On a true
 * square silhouette, which is what these uploads are, it makes no difference at
 * all: it only decides which way a mis-cut file drifts.
 */
.ota-page-hero .ota-page-hero__image {
	display: block;
	inline-size: 100%;
	max-inline-size: none;
	block-size: 100%;
	object-fit: cover;
	object-position: center;

	/*
	 * Left corners only. The frame rounds the slot with `radius/md`, and the
	 * right pair never shows because the column is flush to the edge of the
	 * band, so rounding them is either invisible or, at a viewport wider than
	 * the content, a pair of nicks in an otherwise straight edge.
	 */
	border-start-start-radius: var(--oph-radius);
	border-end-start-radius: var(--oph-radius);
}

/*
 * The brand mark, when there is no feature image. Lifted wholesale from
 * `.ota-footer__mark` in ota-footer.css, which carries the full reasoning:
 * a masked element rather than an <img> so the overlay role can paint it,
 * anchored to a corner and then pushed back out past it by a share of its own
 * size so it keeps bleeding correctly at any band height.
 *
 * BOTTOM RIGHT HERE, BOTTOM LEFT IN THE FOOTER, and that is the whole point of
 * the pair. The footer's moved left in the same change so a page carrying both
 * reads as one composition rather than two graphics stacked in the same corner.
 * No mirroring here: this is the corner the mark was drawn to bleed off.
 *
 * It replaces the frame's ochre staircase, which was too heavy in practice. The
 * mark is the same overlay role, so it still changes colour per theme.
 *
 * Positioned against the band, not sat in the media column: the column is a
 * 464 rectangle and would crop the bleed square off.
 */
.ota-page-hero__mark {
	/*
	 * Named here as well as on the modifier so a mark that somehow arrives with
	 * no shape class still draws something rather than an invisible box.
	 */
	--oph-mark-shape: url("../img/page-hero/ring.svg");

	position: absolute;
	inset-inline-end: 0;
	bottom: 0;

	/*
	 * SIZED OFF THE BAND'S HEIGHT, NOT ITS WIDTH, and this is the important
	 * line. The footer's mark takes a share of the width because a footer is
	 * short and wide and the mark is drawn against a 1440 artboard. Copying that
	 * here looked right at 1440 and fell apart above it: the band's height is
	 * pinned near 464 while its width is the whole viewport, so a percentage of
	 * the width kept growing until the shape was taller than the band it sat in
	 * and lost its top and bottom to the crop. Off the height it stays the same
	 * size relative to the band at every viewport width.
	 *
	 * Square, unlike the footer's 583.487 / 564.91. These six are 115px tiles
	 * rather than a composed corner graphic, and `contain` keeps each one's own
	 * ratio inside the box, so the staircase's stray 115x117 letterboxes by a
	 * pixel instead of being stretched.
	 */
	/*
	 * The two numbers below come off the "Hero treatments" component, node
	 * 2838:19635, which draws five of these against a 464 band. Measured as
	 * shares of the band's height:
	 *
	 *   variant   shape size   top gap   bleed
	 *   2         123.6%       29.6%     43.0%
	 *   3         155.4%       20.6%     47.4%
	 *   4         186.2%        0.0%     46.3%
	 *   5         156.5%       27.0%     48.1%
	 *   6         114.5%       27.4%     36.6%
	 *
	 * No two agree, so this is a range rather than a spec and these values sit
	 * in the middle of it: 135% lands between variant 6's 115 and variant 3's
	 * 155, and a 45% bleed between variant 6's 37 and variant 5's 48. That puts
	 * the top gap at 26%, inside the 21 to 30 the four non-zero variants use.
	 *
	 * Bleed is a share of the SHAPE, size a share of the BAND, which is why the
	 * two percentages are not comparable and why the top gap falls out of them
	 * rather than being set: gap = size - bleed - 100.
	 *
	 * The frame's sixth treatment, `Property 1=Default`, is deliberately not in
	 * that table. It sits flush in the corner at 73.5% with no bleed at all,
	 * which is the original staircase placement rather than one of these.
	 */
	block-size: 135%;
	inline-size: auto;
	aspect-ratio: 1 / 1;
	transform: translate(45%, 45%);
	background-color: var(--oph-mark);
	-webkit-mask: var(--oph-mark-shape) no-repeat center / contain;
	mask: var(--oph-mark-shape) no-repeat center / contain;

	/* Behind the copy, which reaches across it on a wide viewport. */
	z-index: 0;
	pointer-events: none;
}

/*
 * The six. Shape only: every one of them paints from `--oph-mark`, so they all
 * follow the page's overlay role and none of them carries a colour of its own.
 * The files are Figma exports with their ochre baked in, which is irrelevant
 * through a mask because only their alpha is read.
 *
 * Keep this list in step with `Page_Hero::MARKS`. The PHP picks the class for
 * the no-script case and the inline script re-picks on top of it; both index
 * the same six names.
 */
/*
 * THE STAIRCASE IS THE ONE SHAPE THE FRAME PLACES EXACTLY, so it does not take
 * the averaged treatment above. `Property 1=Default` on node 2838:19635 sits it
 * flush in the corner with no bleed at all: a 275.35 shape against that
 * variant's 374.4 band, which is 73.5%, hard against the right and bottom edges.
 *
 * That reads as a step running into the corner rather than a shape sliding off
 * it, which is the whole point of the silhouette, and it is why the averaged
 * 135% and 45% looked so wrong here: bleeding a staircase off a corner cuts the
 * steps in half and leaves an unreadable wedge.
 */
.ota-page-hero__mark--staircase {
	--oph-mark-shape: url("../img/page-hero/staircase.svg");

	block-size: 73.5%;

	/*
	 * THE ONLY FILE THAT IS NOT SQUARE, at 115x117, and the square box above put
	 * a 3px strip of background down its right hand side: `contain` fits by the
	 * longer edge, which here is the height, and centres what is left over.
	 * Measured, not guessed: a pixel scan of the painted band found the other
	 * five reaching column 1439 of 1439 and this one stopping at 1436. Matching
	 * the box to the file makes `contain` fill it exactly.
	 */
	aspect-ratio: 115 / 117;

	/*
	 * Cropped on the right only, which squares off the corner. The path starts
	 * its bottom right curve at x=110.3 of 115, so 4.5% is just past the radius
	 * and the corner comes out hard rather than rounded.
	 *
	 * X only, deliberately. Bleeding the bottom as well would take the rounded
	 * bottom LEFT corner with it, and the frame keeps that one: the shape is
	 * meant to sit on the band's floor and run off its right hand edge, not
	 * slide out of the corner diagonally like the other five.
	 */
	transform: translateX(4.5%);
}

.ota-page-hero__mark--cog {
	--oph-mark-shape: url("../img/page-hero/cog.svg");
}

/*
 * THE BRAND MARK MATCHES THE FOOTER'S DRAWN SIZE, not its proportions. Those
 * are two different things here and the difference is not small: the footer is
 * 622 tall and this band is 464, so copying the footer's 93.8% share gave 435
 * against the footer's 583, and the same graphic came out a third smaller a
 * screen apart. Matched on pixels instead.
 *
 * 583 and 273 are the footer's own numbers at 1440, measured off the live
 * element. Expressed against the band's height rather than the viewport's width
 * so this one holds its size on a wide monitor, which the footer's does not:
 * see the note in ota-footer.css. 125.6% of 464 is 583.
 */
.ota-page-hero__mark--ring {
	--oph-mark-shape: url("../img/page-hero/ring.svg");

	block-size: 125.6%;
	transform: translate(46.8%, 46.8%);
}

.ota-page-hero__mark--circle {
	--oph-mark-shape: url("../img/page-hero/circle.svg");
}

.ota-page-hero__mark--donut {
	--oph-mark-shape: url("../img/page-hero/donut.svg");
}

.ota-page-hero__mark--octagon {
	--oph-mark-shape: url("../img/page-hero/octagon.svg");
}

/*
 * Stacked below 1024, which is where the 464 slot leaves the body column too
 * narrow for the 810px intro measure to mean anything. The square carries over
 * from the base rule against a full bleed width, and the mobile frame draws the
 * media at very close to square anyway, so nothing here needs a second ratio.
 */
@media (max-width: 1024px) {
	/*
	 * The floor comes off with the row layout. Stacked, the band is the copy
	 * plus the media one under the other, and a 464 minimum on top of that just
	 * adds dead background under a short page.
	 */
	.ota-page-hero {
		flex-direction: column;
		min-block-size: 0;
	}

	/*
	 * `flex: 1 1 0` is a row instruction. Once the band turns into a column it
	 * means "basis zero, then grow", which in an auto height container leaves
	 * the body taller than its content and drops a band of dead background
	 * between the intro and the image. Stacked, the body should just be as tall
	 * as what is in it.
	 */
	.ota-page-hero__body {
		flex: 0 0 auto;
		padding-inline: 40px;
	}

	/*
	 * Stacked there is no band height to stretch to, so the square comes back
	 * as a ratio against the width. The mobile frame draws the media at very
	 * close to square anyway.
	 *
	 * CAPPED AT THE SLOT WIDTH, AND THAT CAP IS THE WHOLE RULE. A square that
	 * takes the full bleed is only sane while the viewport is phone width: at
	 * 100% of 768 it is a 768px tall square, so the band went from 464px at
	 * 1025 to 1060px at 768 and 1321px at 1024, a whole screen of hero on a
	 * tablet off a one pixel change in width. Measured, not theorised.
	 *
	 * The cap cannot be a height with `cover`, which is the obvious other
	 * answer: these uploads are transparent PNGs cut to a silhouette, so
	 * cropping one does not show less of a photograph, it lops the top off the
	 * shape. Limiting the width keeps the whole silhouette and lets the square
	 * follow from it.
	 *
	 * Pinned to the trailing edge rather than centred, so between 465 and 1024
	 * it sits where the desktop row puts it and keeps bleeding off the same
	 * edge. Below the cap `min()` gives 100% and the pin is a no-op, which is
	 * the full bleed the mobile frame draws.
	 */
	.ota-page-hero__media {
		flex: 0 0 auto;
		align-self: auto;
		inline-size: min(100%, var(--oph-media));
		margin-inline-start: auto;
		aspect-ratio: 1 / 1;
	}

	/*
	 * THE MARK SHRINKS TO A CORNER TAG AND THE COPY STOPS SHORT OF IT. Sized off
	 * the band's height, the base rule gives a 570px square on a 390px band, and
	 * with no photo to push the copy aside it lands straight under the title and
	 * intro. Every theme paints it in the overlay role, so the damage is whichever
	 * pairing is worst: teal's mint shape under ash text measured 1.01:1 on the
	 * intro, mint and aqua 1.63, ochre 2.43 on the title.
	 *
	 * Below 1024 only, so the desktop band is untouched. Sized off the viewport
	 * here, which is the right base now the band is stacked and has no fixed
	 * height to follow, and capped so a tablet does not get a shape bigger than
	 * the phone one by a factor of two.
	 *
	 * The bottom padding the copy already has is reused: `max()` reserves only
	 * what the shape needs beyond it, so the band grows by the difference rather
	 * than by the whole shape. Two classes on the selectors outrank the
	 * per-shape modifiers above, which set the size and bleed at one class each.
	 *
	 * It still bleeds off the corner the way the desktop mark does, so it reads
	 * as the same cropped brand shape rather than a sticker: pushed out by a
	 * third of its size on both axes, the band's clip cuts it, and only the part
	 * left inside the band is reserved under the copy, plus 16px so the shape
	 * never touches the last line.
	 */
	.ota-page-hero--no-media {
		--oph-mark-size: min(45vw, 260px);
		--oph-mark-bleed: 33%;
		--oph-mark-reserve: calc(var(--oph-mark-size) * 0.67 + 16px);
	}

	.ota-page-hero--no-media .ota-page-hero__body {
		padding-block-end: max(var(--oph-pad), var(--oph-mark-reserve));
	}

	.ota-page-hero--no-media .ota-page-hero__mark {
		block-size: var(--oph-mark-size);
		transform: translate(var(--oph-mark-bleed), var(--oph-mark-bleed));
	}

	/*
	 * The staircase is drawn to sit square in the corner and never bleeds, on
	 * desktop either, so it keeps its own smaller size and reserves all of it.
	 * `:has()` follows the inline script that re-picks the shape on load.
	 */
	.ota-page-hero--no-media:has(.ota-page-hero__mark--staircase) {
		--oph-mark-size: min(30vw, 180px);
		--oph-mark-reserve: calc(var(--oph-mark-size) + 16px);
	}

	/* Keeps the bleed off the right edge that squares the step into the corner. */
	.ota-page-hero--no-media .ota-page-hero__mark--staircase {
		transform: translateX(4.5%);
	}
}

@media (max-width: 600px) {
	/*
	 * Scaled with the base rather than left where it was. 32 was a third off a
	 * 48px desktop gutter; against 92 it reads as no padding at all on a phone,
	 * where the band is the first thing under the header.
	 */
	.ota-page-hero__body {
		gap: 16px;
		padding-block: 56px;
		padding-inline: 20px;
	}

	/* The 1024 rule's `max()` against this narrower padding, which it outranks. */
	.ota-page-hero--no-media .ota-page-hero__body {
		padding-block-end: max(56px, var(--oph-mark-reserve));
	}

	.ota-page-hero__intro {
		font-size: 16px;
		line-height: 25.6px;
	}
}
