1. Footer still references Quantify — will silently fall back after this PR (bug)

docs/site/src/theme/Footer/index.tsx:47:
<span style={{fontFamily: "'Quantify', sans-serif", fontWeight: 700, ...}}>erigon.tech
With the @font-face removed and Quantify.woff2 deleted, this brand text will fall back to whatever the browser resolves sans-serif to — different from both Quantify and the navbar's
new Montserrat 800. The PR description says "No change to headings, body, or any other element", but the footer brand text will visibly change.

The PR reasoning explicitly calls out that "Quantify is the brand font on erigon.tech (logo wordmark only, two occurrences: header + footer)" — so the same fix that's being applied to
the navbar should be applied to the footer in the same PR. Otherwise the navbar and footer brand text will diverge.

2. docs/site/DESIGN-SPEC.md is now stale

Four references contradict the new state:

• Line 119: "Font: Quantify, weight 700, letter-spacing 0.04em" (navbar title)
• Line 159: "Brand text: Quantify font, weight 700" (footer brand)
• Line 272: @import url('https://fonts.cdnfonts.com/css/quantify'); (already stale from #20991, but worth fixing here)
• Line 280: | Navbar title, footer brand | Quantify | 700 |

The doc was written as a brand spec, so leaving it describing a font that no longer ships is confusing for future contributors.

Pull request overview

Updates the docs site branding typography by removing the custom Quantify font usage and switching the navbar title (and footer brand text) to the already self-hosted Montserrat ExtraBold (800), aligning the docs UI with the stated brand guidelines.

Changes:

• Removed the @font-face for Quantify and updated .navbar__title to Montserrat weight 800.
• Updated the footer “erigon.tech” brand text to use Montserrat weight 800.
• Adjusted the design spec to reflect the new navbar/footer brand font choice (but see documentation note below).

Reviewed changes

Copilot reviewed 3 out of 4 changed files in this pull request and generated 1 comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Pull request overview

Copilot reviewed 3 out of 4 changed files in this pull request and generated no new comments.

Pull request overview

Copilot reviewed 3 out of 4 changed files in this pull request and generated no new comments.

Hmm. We can keep Quantify because we have the licence now, right?

Overview

Three independent changes bundled into one PR:

1. Brand font realignment — Quantify → Montserrat 800 for navbar + footer (erigon.tech), and deletes Quantify.woff2 since no text on the docs site uses it anymore.
2. Typography polish in custom.css — letter-spacing on h1/h2, h6 as a small uppercase label, tighter heading line-height, font smoothing, vertical rhythm, scroll-margin-top.
3. Install-page scroll behavior rewrite — replaces the toggle-event-listener-then-scroll flow with a "wait for stable layout, then resolve target section heading, then smooth-scroll" flow that also closes sibling <details> and updates location.hash.

Approving — direction is right, brand fix is correct, all four stale DESIGN-SPEC.md references caught after the prior round. A few suggestions below, none blocking.

What's good

• Self-hosted fonts only. Replaces two Google/CDN @import lines with explicit @font-face declarations. Removes a cross-origin runtime dependency and a referrer leak.
• history.replaceState not pushState for the hash update — correct UX choice; clicking install-method tabs shouldn't pollute back-button history.
• 6-frame stability + 90-frame cap in waitForStableLayout — sensible defensive pattern: bounded worst case (~1.5s @ 60Hz) and avoids the previous race between toggle animation and smooth-scroll.
• scroll-margin-top: calc(var(--ifm-navbar-height) + 1rem) on all .markdown headings — the right fix for in-page anchor offset under a sticky navbar.

Suggestions (non-blocking)

1. if (e) e.preventDefault() in the new go(id, e) is dead code

The go function signature gained an e parameter, but all call sites are unchanged and still look like:

<a href=\"#install-prebuilt\" onClick={e => { e.preventDefault(); go('install-prebuilt'); }}>

go is invoked with a single argument, so e is always `undefined` inside go and the new guard never fires. Either drop the second param, or simplify call sites to onClick={e => go('install-prebuilt', e)} and rely on go to call preventDefault. Don't ship both forms.

2. summary.click() for opening/closing <details> is roundabout

const s = d.querySelector('summary');
if (s) s.click();
else d.open = false;

Direct d.open = false / el.open = true produces the same visible result, fires the toggle event, and is one less thing to reason about. A synthetic click runs through any other listeners on <summary> and depends on the browser's default-action toggle behavior — indirect for no apparent gain.

3. h6 visual reassignment changes existing semantics

Repurposing h6 as a small uppercase muted label is a stylistic call, but any existing MDX page that uses ###### as a real sub-sub-heading will silently re-render as uppercase muted text. Worth either grepping the docs to confirm no collateral damage, or documenting the new h6 role explicitly in DESIGN-SPEC.md (the spec currently lists h1–h6 as one row at Montserrat 800).

4. Typography polish not reflected in DESIGN-SPEC.md

The PR fixes the spec for the Quantify→Montserrat changes but adds new rules the spec doesn't describe:

• -0.015em / -0.01em letter-spacing on h1/h2
• --ifm-heading-line-height: 1.2
• h6 as small uppercase label
• vertical rhythm above .markdown h2/h3
• scroll-margin-top on markdown headings
• text-rendering: optimizeLegibility + font-smoothing on <html>

If DESIGN-SPEC.md is the brand source-of-truth, a "Typography rhythm" subsection that mirrors what landed in custom.css would keep it complete.

5. text-rendering: optimizeLegibility on <html> — minor perf caveat

Forces ligatures + kerning across the entire document. There's a long-standing known cost on very text-heavy pages and on older Android browsers where it triggers a slower text path. Consider scoping to headings (h1,h2,h3,h4,h5,h6 { text-rendering: optimizeLegibility; }) where the visual gain is real, rather than globally on <html>.

6. resolveTarget mapping is fragile

sectionForMethod is hardcoded against three Docusaurus-generated section ids (linuxmacos, windows, all-operating-systems). If the docs author later renames ### Linux/macOS to ### Linux / macOS, the slug changes (linux--macos or similar), the mapping breaks silently, and we fall through to the previous-sibling walk — at which point we're scrolling to whatever heading happens to precede the details element. Two ways to harden:

• Add a runtime warning when sectionForMethod[id] resolves to a non-existent element.
• Better, walk up the DOM from the <details> until the first heading is found, removing the hardcoded mapping entirely.

Fine for now since it'll be discovered visually, but worth acknowledging as a coupling between content and JS.

7. Add font-display: swap to the new @font-face blocks

The previous Quantify rule had font-display: swap explicitly; the four new @font-face declarations omit it and rely on the browser default (auto). Adding swap keeps FOUT behavior consistent and explicit.

8. Scope

Original change was the Quantify→Montserrat brand fix; typography polish and install-page rewrite rode along. Each is independently reviewable. Bundling means a regression in one area shares a single git revert with the others. Not blocking, but worth a thought for future PRs in this area.

9. CI

docs-site / build is red; the author flagged it as a pre-existing SidebarsConfig TS error fixed by another PR. The actual fix appears to be #21083 (docs: fix SidebarsConfig TypeScript import for Docusaurus 3.10 compatibility), not #21074 (which is the llms.txt regen). Worth confirming the merge order with that PR.