Repository navigation
chore: standardise website/ on npm and delete the duplicate lockfiles - #365
Conversation
…IntersectMBO#337) - Delete website/yarn.lock and the stray root package-lock.json, leaving website/package-lock.json as the only tracked lockfile — the duplicate was why Dependabot double-counted the braces advisory (alerts 259/260) and reported node-forge from a lockfile CI never installs (alert 261). - Remove packageManager (yarn) and the yarn-only resolutions block from website/package.json; overrides already carries the identical 32 pins for npm (verified key-for-key), so nothing is lost. - Docs now describe npm only: website/README.md (npm install, npm run start:dev, npm run build; deployment section now points at the deploy workflow instead of manual yarn deploy to gh-pages), CONTRIBUTING.md (npm, npm run start:dev), and website/SECURITY.md (commit package.json + package-lock.json together). - Guard in pr-build-website.yml fails the PR build if any yarn.lock / pnpm-lock.yaml or a root package-lock.json is tracked again, and the workflow now also triggers on lockfile paths outside website/. Acceptance criteria for IntersectMBO#337: only website/package-lock.json tracked; no packageManager/resolutions; README + CONTRIBUTING npm-only; guard fails on a reintroduced lockfile (tested both ways locally). The Dependabot alert closure follows the post-merge scan.
cba7397 to
4e58a7d
Compare
|
Hey Kyle, Going to review this before merging. |
|
Sounds good
…On Sun, Oct 18, 2026, 10:45 AM Ian Njuguna ***@***.***> wrote:
*IanoNjuguna* left a comment (IntersectMBO/developer-experience#365)
<#365 (comment)>
Hey Kyle, Going to review this before merging.
—
Reply to this email directly, view it on GitHub
<#365?email_source=notifications&email_token=CMAPYQEHLBRLGQNUR3VFFX35UTQQBA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTMMRVGE4DIOJSG432M4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2KYZTPN52GK4S7MNWGSY3L#issuecomment-6251849277>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/CMAPYQCL5JVWHY5JSMC3IPT5UTQQBAVCNFSNUABFKJSXA33TNF2G64TZHM4TANBYGUZDOOJWHNEXG43VMU5TKOJQHA3TCMBUGUYKC5QC>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Is this pull request abandoned? It had no activity for 21 days. It has been marked stale and will be closed in 7 days if no further activity occurs. Add a comment or push a commit to keep it open. See TRIAGE.md for our triage cadence. |
|
Ping @IanoNjuguna @danbaruka |
|
Hi @Kshot3000, This is the right fix for #337. resolutions and overrides are the same 32 pins on main, so dropping resolutions and website/yarn.lock loses nothing. |
Conflicts: - CONTRIBUTING.md: keep upstream's Node 22.23.4 prerequisite text, keep the npm-only package manager line from this branch. - website/SECURITY.md: keep the single-lockfile commit rule from this branch, keep upstream's new Node runtime section, and correct that section's claim that packageManager still declares Yarn 1.22.22.
The pin narrative still described `resolutions` as a live, kept-in-sync block. With yarn gone from the repo, `overrides` is the only pin mechanism CI applies.
Resolutions: - website/yarn.lock: modified upstream, deleted here. Deletion wins, npm is the only installer. - website/package.json: upstream's November override values are kept as-is (33 pins, verified identical to the resolutions block they replace); packageManager and resolutions stay removed. - website/SECURITY.md: upstream's November pin table and residual finding are kept; the two sentences that described resolutions as live, and the one that pointed readers at a second block, now describe overrides only.
The skill blocked exactly the change this branch makes, and its preferred workflow (yarn add) writes yarn.lock and never package-lock.json, so following it produced a PR that fails npm ci. It now enforces the standard the workflows actually run: one lockfile, npm ci, pins in overrides.
|
Conflicts resolved on main and pushed to this branch. How each one was resolved:
Those two hand edits on main this morning are the case for this PR: every dependency change was being written into two lockfile formats by hand, and one of them was not installable ( One addition, since it was the last thing contradicting the change: Verified on the merged tree: guard passes, @Kshot3000 your commit is untouched and still authored by you. The merge commit and the skill edit are mine, so if you want either worded differently, say so. |
IanoNjuguna
left a comment
There was a problem hiding this comment.
Approving. website/ is npm from here on: one lockfile (website/package-lock.json), pins in overrides
|
Im just happy to help Cardano and Midnight in any way I can. Thanks guys.
…On Wed, Nov 11, 2026, 2:10 AM Ian Njuguna ***@***.***> wrote:
Merged #365
<#365> into main.
—
Reply to this email directly, view it on GitHub
<#365?email_source=notifications&email_token=CMAPYQBOSEEJH7BWQK665SD5YQNGNA5CNFSNUABQM5UWIORPF5TWS5BNNB2WEL2JONZXKZKFOZSW45CON52GSZTJMNQXI2LPNYXTGNJXG42TEOJZGYZTNJTSMVQXG33OU5WWK3TUNFXW5JLFOZSW45FMMZXW65DFOJPWG3DJMNVQ#event-35775299636>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/CMAPYQGUZUQCVFQ6PDNKJI35YQNGNAVCNFSNUABFKJSXA33TNF2G64TZHM4TANBYGUZDOOJWHNEXG43VMU5TKOJQHA3TCMBUGUYKC5QC>
.
You are receiving this because you were mentioned.Message ID:
***@***.***
.com>
|
Fixes #337 — standardises
website/on npm and deletes the duplicate lockfiles.What changed
website/yarn.lockand the stray rootpackage-lock.json(an emptypackages: {}lock with no matching manifest).website/package-lock.jsonis now the only tracked lockfile — the one CI actually installs from. This is the duplication that double-counted the braces advisory as alerts 259 + 260 and produced the false node-forge alert 261 from a lockfile CI never installs; both should close on the post-merge Dependabot scan.website/package.json: removed the"packageManager": "yarn@…"declaration and the yarn-onlyresolutionsblock. Nothing is lost: I verifiedresolutionsand the npmoverridesblock are identical — same 32 pins, key for key — before deleting.website/README.md—npm install,npm run start:dev(with a note thatnpm startis a full production build + serve, not the dev server),npm run build,npm run audit. The deployment section no longer tells contributors toyarn deploytogh-pagesby hand; it points at the deploy workflow that runs on push tomain.CONTRIBUTING.md— prerequisites are npm (not "npm or yarn"), and the local-server step isnpm run start:dev.website/SECURITY.md— one line in the same spirit: commitpackage.jsonandpackage-lock.jsontogether (it previously told contributors to commityarn.locktoo, which would reintroduce the duplicate).pr-build-website.ymlgains a step that fails the PR build if any trackedyarn.lock/pnpm-lock.yaml, or a rootpackage-lock.json, exists — and the workflow's path triggers now include lockfile paths outsidewebsite/**so a root-level lockfile PR runs the gate too.Verification
website/yarn.lockforce-added back → guard reports it (fail).website/package.jsonparses as valid JSON after the deletions; the diff is deletions-only there (35 lines).pr-build-websitecheck (npm ci+npm run buildfrom the surviving lockfile).Notes
CONTRIBUTING.md, in adjacent lines (mine: the package-manager prerequisite and the start command; chore: pin the Node version for contributors #342: the Node version line). If chore: pin the Node version for contributors #342 merges first, this branch may need a trivial rebase — happy to do it..agents/skills/harun-review/SKILL.md, which still describes yarn as the source of truth — that one reads like a maintainer-owned review skill, so flagging it here rather than rewriting it in this PR.Built as an independent Cardano community contributor — not affiliated with the Cardano Foundation, IOG or Intersect.
@Kshot9000 · https://x.com/kshot9000
ADA: addr1q8hnl6vl5a6k3rw3n5g3jtte696zcl76kfatzv7gpswa9r0dj7fma6klq55y4ffm7tf0em09udnyhuk4ah92pl5x9jpqjae44v
cc @cardano-foundation @IntersectMBO