tudor/school_compare
1
0
Fork 0
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity
681 Commits 74 Branches 0 Tags
39e5c1127470716e1bad9e470edb3e7adfa915f1
Commit Graph
4 Commits
Author SHA1 Message Date
Tudor 39e5c11274 fix(brand): match the below-average tint to its own hue, and finish the casing
PR Checks / Frontend Typecheck + Tests (pull_request) Successful in 1m2s
Details
PR Checks / Backend Smoke (pull_request) Successful in 6s
Details
PR Checks / Build Backend (no push) (pull_request) Successful in 10s
Details
PR Checks / Build Frontend (no push) (pull_request) Successful in 43s
Details
PR Checks / Build Pipeline (no push) (pull_request) Successful in 10s
Details
PR Checks / AI Code Review (Claude) (pull_request) Successful in 3m19s
Details 
Three review findings.

--status-below-bg carried the raw --coral tuple (249,115,96) instead of the
hue it belongs to (#A9481F → 169,72,31), so "below average" chips rendered a
coral-pink wash behind terracotta text — the one solid/tint pair in the file
whose hues disagreed. Alpha drops 0.15 → 0.12 to match --status-above-bg, so
the two chips read as equal weight; left at 0.15 the darker terracotta tint
takes the label to 4.45:1 on Warm White, under the 4.5 floor. At 0.12 it is
4.86:1 on cards and 4.67:1 on Warm White.

Auditing the whole file for the same class of error turned up one more:
--action-bg was also a --coral tint while --action is the darkened #BE3C27.
It is unused today and the dark-theme value was already correct, but it is
aligned so a future coral ghost button gets a matching label and ground. The
five phase pairs also "mismatch" by the same test and are deliberately left
alone: those pair a pastel ground (sage, sky, mustard, lavender) with a deeper
text hue, which is exactly what the guideline's Chips/Tags show.

The lowercase wordmark had only reached the visual chrome. Every user-visible
occurrence now matches: page title and template, appleWebApp title, authors,
OpenGraph and Twitter titles and siteName, the school page's siteName, the PWA
manifest name and short_name, the footer copyright line, the nav's aria-label,
the Web Share API strings, and the editorial paragraph. Comments in lib/*.ts
and AdmissionsView still say "SchoolCompare"; they describe the codebase rather
than the mark, so they are left to keep this diff to the reported problem.

The comment explaining why the font classes belong on <html> still described
--font-schibsted, which no longer exists. It now names --font-manrope and
--font-inter, and records that the e2e typeface assertion is the only guard —
this failure mode throws nothing and fails no unit test.

Verified: tsc clean, 159/159 tests, build green, and the rendered title is now
"School Admissions Guide | schoolcompare". Status and phase labels re-checked
on their own composited tints over both light grounds — all clear AA.
 2026-08-07 18:33:30 +01:00
TudorandClaude Opus 4.8 13253116c2 fix(home): correct remaining KS2/Ofsted terminology on the landing page
PR Checks / Frontend Typecheck + Tests (pull_request) Successful in 1m1s
Details
PR Checks / Backend Smoke (pull_request) Successful in 7s
Details
PR Checks / Build Backend (no push) (pull_request) Successful in 10s
Details
PR Checks / Build Frontend (no push) (pull_request) Successful in 41s
Details
PR Checks / Build Pipeline (no push) (pull_request) Successful in 9s
Details
PR Checks / AI Code Review (Claude) (pull_request) Successful in 1m35s
Details 
Follow-up to the accuracy pass — the medium-severity wording issues
from the same expert review:

- KS2 "Exceeding" isn't a DfE term. The above-expected level is the
  "higher standard" for the reading and maths tests and "greater depth"
  for teacher-assessed writing. Labelled per subject, and brought the
  illustrative reading value down from an implausible 73% (national
  higher standard is ~29%) to a realistic strong-school figure.
- Attributed inspection outcomes to Ofsted, not only the DfE (Ofsted is
  a separate non-ministerial department).
- Hero: "2026/2027 admissions results" -> "the 2026/27 admissions
  round" (admissions produce offers, not results, and we don't hold
  per-school offer outcomes).

Tightened .miniRowHead so the longer "Greater depth" label stays on one
line and the three subject columns keep their bars aligned (the visual
is desktop-only; hidden on mobile).

Verified with tsc --noEmit and next build.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0146VHeLAWjDVE2B5uU67jCB
 2026-07-21 09:07:23 +01:00
TudorandClaude Opus 4.8 1b8adb91ee fix(home): correct education/Ofsted claims on the landing page
PR Checks / Frontend Typecheck + Tests (pull_request) Successful in 1m3s
Details
PR Checks / Backend Smoke (pull_request) Successful in 7s
Details
PR Checks / Build Backend (no push) (pull_request) Successful in 12s
Details
PR Checks / Build Frontend (no push) (pull_request) Successful in 42s
Details
PR Checks / Build Pipeline (no push) (pull_request) Successful in 10s
Details
PR Checks / AI Code Review (Claude) (pull_request) Successful in 37s
Details 
Homepage marketing content had drifted from both the current
education regime and the app's own school pages. An expert accuracy
review surfaced five issues (3 critical, 2 high):

- Ofsted card showed a single "OUTSTANDING" word-grade. Overall grades
  were abolished (Sept 2024) and inspections now produce a report card
  on a five-point scale (from Nov 2025). Rebuilt the card as a report
  card with per-area ratings + stand-alone safeguarding, matching what
  CompareOfsted actually renders on a school page.
- Attainment 8 "National avg 50.2" was wrong; the real 2024/25 figure
  is 45.9. Corrected the benchmark and the illustrative school value.
- Compare mock featured a KS2 "Reading progress" score, but KS2
  progress isn't published for 2023/24 or 2024/25 (no post-COVID KS1
  baseline). Swapped for Higher standard (RWM), which is published.
- Editorial called it "UK school performance data"; the data is
  England-only. Now matches the hero and page metadata.
- Softened the "40-page inspection report" exaggeration.

Landing-page copy isn't covered by e2e journeys, so no test changes.
Verified with tsc --noEmit and next build.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0146VHeLAWjDVE2B5uU67jCB
 2026-07-21 09:00:02 +01:00
Tudor SitaruandClaude Opus 4.7 62eeee5f7c perf: cache aggressively and trim client bundle
Build and Push Docker Images / Build Backend (FastAPI) (push) Successful in 1m1s
Details
Build and Push Docker Images / Build Frontend (Next.js) (push) Successful in 53s
Details
Build and Push Docker Images / Build Pipeline (Meltano + dbt + Airflow) (push) Successful in 2m4s
Details
Build and Push Docker Images / Trigger Portainer Update (push) Successful in 1s
Details 
Frontend
- Dynamic-import Chart.js components on detail/compare views so Chart.js
  no longer ships in initial JS.
- Drop force-dynamic on home, compare, rankings so internal data fetches
  reuse Next.js's per-call revalidate cache.
- Switch /school/[slug] to ISR with a 7-day revalidate window (school
  data updates annually).
- Preconnect to analytics + postcodes.io; remove redundant defer on the
  Umami Script tag (afterInteractive already covers it).
- Bump images.minimumCacheTTL to 1 year.
- Extract HowItWorks and Editorial sections as server components passed
  to HomeView via slot props so their JSX stays out of the client bundle.

Backend
- Add GZipMiddleware (min 512 bytes).
- Add CacheAndETagMiddleware: per-path Cache-Control with long s-maxage
  + stale-while-revalidate, ETag generation, and 304 on If-None-Match.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
 2026-06-02 13:46:45 +01:00