chore: #1 pnpm workspace, ESLint, Prettier, TypeScript, minimal client #32

erikhfj · 2026-09-25T18:33:27Z

What this does

Sets up the repo tooling for #1 so the four of us can work in parallel and the same quality gates run locally and in CI.

Piece	Choice
Workspace	`pnpm` with `client/` only. `server/` and `e2e/` join in M2 and M4 rather than existing as empty packages now.
TypeScript	`tsconfig.base.json` with `strict` and `noUncheckedIndexedAccess`
Lint	ESLint 9 flat config - typescript-eslint, react-hooks, jsx-a11y, react-refresh
Format	Prettier, applied last via eslint-config-prettier so the two never fight
Scripts	`dev`, `build`, `lint`, `lint:fix`, `format`, `format:check`, `typecheck`, `test`, `test:watch`
Docs	`AGENTS.md` with the architecture rules we agreed on

The client is a skeleton only - it renders the product name, a skip link and a main landmark. Its whole job is to prove the gates are green before #9 replaces it with the mock-data search interface.

What was done

pnpm workspace (pnpm-workspace.yaml) declaring client/
tsconfig.base.json extended by the client
eslint.config.js - TypeScript rules for everything, React + a11y rules scoped to client/**
.prettierrc.json / .prettierignore
.gitignore including *.tsbuildinfo
Root package.json with the scripts above
AGENTS.md - the rules that make the graded requirements work (SQL-side sort/filter, keyset pagination, DataLoaders, simulated users, plain CSS, accessibility non-negotiable, search state in the URL)
client/ minimal React + Vite + Vitest setup

What was tested

Ran from a clean pnpm install:

pnpm format:check   ✅  all files formatted
pnpm lint           ✅  0 errors, 0 warnings
pnpm typecheck      ✅  tsc -b --noEmit clean
pnpm test           ✅  3 tests passed
pnpm build          ✅  220 kB JS (68.8 kB gzip), 1.1 kB CSS

The three tests cover:

the heading and tagline render
the skip link is present and points at #main
a main landmark exists

Two accessibility details are already wired in so #9 inherits them rather than retrofitting: the skip link, and :focus-visible outlines. prefers-color-scheme and prefers-reduced-motion are in from the start.

Notes on choices

noUncheckedIndexedAccess is annoying to write against. It is deliberate - it catches the "index might be undefined" class of bug at compile time, which is exactly the kind that surfaces later as a blank row in the results list.
0 warnings, not "nearly clean". pnpm lint is what CI runs, so the bar has to be enforceable rather than aspirational.
Plain CSS, no framework. Smaller bundle for bærekraft, and full control over focus rings and contrast for tilgjengelighet. Same call as Project 1.
Only client/ in the workspace. Empty server//e2e/ packages would show the eventual shape but add nothing to review today.

Closes

Closes #1.

Sets up the tooling so four people can work in parallel and the same quality gates run locally and in CI. - pnpm workspace with client/ only. server/ and e2e/ join in M2 and M4 rather than existing as empty packages now. - tsconfig.base.json: strict + noUncheckedIndexedAccess. - ESLint 9 flat config (typescript-eslint, react-hooks, jsx-a11y, react-refresh) with eslint-config-prettier applied last. The bar is 0 warnings so CI can enforce it. - Prettier with .prettierrc.json and .prettierignore. - Root scripts: dev, build, lint, format, format:check, typecheck, test. - AGENTS.md documenting the architecture rules we agreed on (SQL-side sort/filter, keyset pagination, DataLoaders, simulated users, plain CSS, accessibility non-negotiable, search state in the URL). - A minimal client that renders the product name, a skip link and a main landmark, purely to prove the gates are green. #9 replaces it with the mock-data search interface. Co-Authored-By: opencode <noreply@opencode.ai>

House style is plain hyphens. No functional change. Co-Authored-By: opencode <noreply@opencode.ai>

rachelks

pnpm format:check fails for me on a clean checkout on Windows and reports formatting issues in 17 files. I checked the line endings and get i/lf w/crlf, and I have core.autocrlf=true, so it looks like an LF/CRLF issue rather than the files actually being formatted incorrectly. Could we add .gitattributes or something similar so the format check works consistently on Windows too?

Found in review of #32: pnpm format:check fails on a clean checkout on Windows for 17 files that are actually formatted correctly. The cause is core.autocrlf=true, the Git for Windows default, which rewrites the checked-out files to CRLF. Prettier then sees different bytes than the ones committed and reports failures that do not exist. Setting `* text=auto eol=lf` means all four of us see the same bytes regardless of platform, so format:check behaves identically everywhere. Shell scripts keep LF for shebangs; batch files get CRLF deliberately; binary assets are excluded from normalization. Renormalizing the tree produced no diff, so the committed files were already LF. This only fixes the checkout side. Reported by rachelks. Co-Authored-By: opencode <noreply@opencode.ai>

erikhfj · 2026-09-28T16:52:20Z

Good catch, and this was a real bug rather than a config nit. Thank you for testing with core.autocrlf=true explicitly, that made the cause obvious.

Cause: Git for Windows defaults to core.autocrlf=true, which rewrites checked-out files to CRLF. Prettier then compares against different bytes than the ones committed and reports 17 failures that do not exist. The files are fine in the repository; the working tree just has different line endings.

Fix: .gitattributes with * text=auto eol=lf. All four of us now see the same bytes regardless of platform, so pnpm format:check behaves identically everywhere. Shell scripts keep LF so shebangs are honoured, batch files get CRLF on purpose, and binary assets are excluded from normalization.

I also ran git add --renormalize . to check whether any committed file needed rewriting. The diff is empty, so this only fixes the checkout side.

Pushed to this branch. Could you re-run pnpm format:check on a fresh Windows checkout and confirm it is clean? If it still complains, the next suspect is Prettier's own endOfLine default and we can pin endOfLine: "lf" in .prettierrc.json as a belt-and-braces measure.

Found by actually verifying rather than assuming: PR #34 targets feat/1-repo-tooling because it is stacked on #32, and the workflow's `pull_request: branches: [main]` filter silently skipped CI for it. The push of 6ab73b6 produced no run at all. Stacked PRs are how we are working (see docs/WORKFLOW.md), so a branch filter on pull_request is wrong. pull_request now has no filter and runs for every PR whatever it targets; push stays limited to main so we only deploy from the mainline. Co-Authored-By: opencode <noreply@opencode.ai>

rachelks

Re-tested with a fresh Windows checkout and core.autocrlf=true. pnpm format:check now passes, so the .gitattributes fix resolves the issue I found.

The workflow is on main for the first time. PR #34 merged into feat/1-repo-tooling (its base at the time) after #32 had already merged that branch into main, so the workflow never reached main and the repo has been running with no CI at all. This brings it across. Content is unchanged from what #34 reviewed and what has been passing on the runner: quality job (format, lint, typecheck, test, build, bundle budget) on every PR and every push to main, deploy job on push to main only. Two fixes that came out of actually running it: - pull_request has no branch filter. A `branches: [main]` filter meant stacked PRs like #34, which targets feat/1-repo-tooling, silently got zero CI runs. - deploy is pinned to [self-hosted, linux, x64, project2]. It writes to /var/www/html/project2 and restarts crate-api, which only exist on it2810-19. quality stays unpinned so any future runner can pick it up. Co-Authored-By: opencode <noreply@opencode.ai>

erikhfj requested a review from a team as a code owner September 25, 2026 18:33

chore: #1 replace em-dashes with hyphens

22ed5d6

House style is plain hyphens. No functional change. Co-Authored-By: opencode <noreply@opencode.ai>

erikhfj mentioned this pull request Sep 25, 2026

chore: #27 CI workflow on the self-hosted runner #34

Merged

rachelks reviewed Sep 27, 2026

View reviewed changes

robertky mentioned this pull request Sep 28, 2026

feat: #2 define Crate data model and GraphQL schema #35

Merged

rachelks approved these changes Sep 28, 2026

View reviewed changes

rachelks merged commit 16ef3cd into main Sep 28, 2026

erikhfj mentioned this pull request Sep 28, 2026

chore: #27 land the CI workflow on main #37

Merged

Uh oh!

chore: #1 pnpm workspace, ESLint, Prettier, TypeScript, minimal client #32

chore: #1 pnpm workspace, ESLint, Prettier, TypeScript, minimal client #32

erikhfj commented Sep 25, 2026 •

edited

Loading

Uh oh!

rachelks left a comment

Uh oh!

erikhfj commented Sep 28, 2026

Uh oh!

rachelks left a comment

Uh oh!

Uh oh!

chore: #1 pnpm workspace, ESLint, Prettier, TypeScript, minimal client #32

chore: #1 pnpm workspace, ESLint, Prettier, TypeScript, minimal client #32

Conversation

erikhfj commented Sep 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this does

What was done

What was tested

Notes on choices

Closes

Uh oh!

rachelks left a comment

Choose a reason for hiding this comment

Uh oh!

erikhfj commented Sep 28, 2026

Uh oh!

rachelks left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

erikhfj commented Sep 25, 2026 •

edited

Loading