chore(release): 0.3.15-unstable.20260831165924 #2369

Workflow file for this run

.github/workflows/code-quality.yml at 05a8df8

	name: Code Quality

	on:
	push:
	# DEFAULT BRANCHES ONLY. `pull_request` below carries every other branch.
	#
	# This was an allow-list of branch prefixes, and that was a gate with a
	# SILENT hole: a branch matching nothing got no CI at all, and its last
	# visible status was whatever it inherited — indistinguishable, on every
	# dashboard, from a branch that passed. Two live examples, both found
	# 2026-08-14: `perf/**` was uncovered in openconnector, where a merge
	# carrying unresolved conflict markers and 84 failing tests was pushed and
	# nothing ran; and `feat/**` was uncovered in openregister, because the
	# list said `feature/**`.
	#
	# The comment that stood here said adding prefixes was not the durable fix,
	# and that the durable fix was to let the pull_request trigger gate it.
	# THIS IS THAT CHANGE.
	#
	# What forced it now: a push to a branch with an open PR ran the SAME 34
	# jobs TWICE on the same commit. `concurrency` cannot dedupe them — the
	# group is suffixed by event name deliberately (.github#540: a
	# default-branch push carries jobs a PR run does not, and a dispatch must
	# not be cancellable by a standing release PR), so the two events sit in
	# different lanes BY DESIGN and both run to completion. Measured fleet-wide
	# 2026-08-25..27, 659 of 2,106 Code Quality runs were that duplicate — 31%
	# of the fleet's most expensive workflow, re-deciding a commit another run
	# was already deciding. The account ceiling is 60 concurrent jobs (Team
	# plan); the fleet was measured at 53 running with 1,528 jobs queued behind
	# them, the oldest run 7 hours old and not yet started.
	#
	# NO BRANCH LOSES ITS FLOOR. merge-hygiene.yml runs on `'**'` — every
	# branch anyone pushes, no prefix list to forget — and it is the check
	# `development` actually requires. That is the smoke alarm; this workflow
	# is the fire brigade and belongs on the PR. Of 668 feature-branch push
	# runs in that window, only NINE were on a branch with no PR run beside
	# them.
	#
	# The default branches STAY: their push runs are not duplicates, they are
	# the only carrier of Coverage Baseline Check, SBOM and Features Extract,
	# none of which run on a pull_request event.
	branches:
	- main
	- development
	pull_request:
	# Spelled out rather than left to the default so the set is reviewable.
	# `synchronize` — a push to an open PR — is the load-bearing one: without
	# it the suite runs once at PR-open and every later commit merges unchecked
	# under the first run's green tick. Cost is bounded by the `concurrency`
	# block below, which cancels the in-flight run for the same head ref.
	types: [opened, reopened, synchronize]
	branches: [main, beta, development]
	workflow_dispatch:

	# Deduplicating a `push` run against the `pull_request` run for the SAME head
	# ref is the point of this block, and for a feature branch it is exactly right:
	# two runs of identical jobs, one of them wasted.
	#
	# It is wrong for `main` and `development`, because the push run there is NOT a
	# duplicate — it carries the push-only "Coverage Baseline Check", which is the
	# whole push side of the coverage ratchet. And those two branches always have an
	# open PR whose `head_ref` IS the branch name: the standing "Release: merge
	# development into beta". `github.head_ref` on that PR run and `github.ref_name`
	# on the push run both render `development`, so both landed in the identical
	# group `quality-development`, and `cancel-in-progress` killed whichever started
	# first — always the push run, by ~9 seconds.
	#
	# Measured on the merge of #1151 (ddd74da4): run 31045325182, event `push`,
	# CANCELLED; run 31045337494, event `pull_request`, survived. Same pattern on
	# 3b9371f1, 75190bca and f2219e09 — four for four, so the push run on
	# `development` has never reached a verdict. "Coverage Baseline Check" therefore
	# reported `skipped` on the surviving run (correctly — it is push-only) while
	# never once executing on the run that could have run it. #1151 enabled a job
	# that could not reach a verdict: a dead gate of the permanently-pending shape,
	# which is invisible because a skipped job renders like a passing one.
	#
	# Suffixing only the default-branch push keeps feature-branch dedup untouched
	# (`quality-feature/x` for both events, exactly as before) and gives the two
	# default branches' push runs a lane of their own.
	concurrency:
	# SUFFIXED BY EVENT NAME, not just by `-push`.
	#
	# The previous expression gave a push on `development` its own lane
	# (`-push`) but left EVERYTHING ELSE sharing `quality-development` — and
	# that is not a quiet lane: `Sync to Beta` keeps a PR open whose head_ref
	# IS `development`, so its run computes the same group and is re-triggered
	# on every merge.
	#
	# A `workflow_dispatch` therefore shared a group with that PR and was
	# cancelled by it. Measured on shillinq 2026-08-21: dispatch 32487948678
	# cancelled by pull_request run 32490160836 (head_branch `development`).
	# A run someone deliberately asked for could essentially never complete.
	#
	# That reaches past ad-hoc verification: the fleet gate-drift sweep
	# (.github#523) dispatches per app with `--ref development`, because
	# `schedule:` cannot choose a branch. Under the old group those runs are
	# cancelled and report neither pass nor fail — and a routine that produces
	# no verdict is indistinguishable from one that never ran.
	#
	# This is hermiq's form, already live there. Pull requests keep the bare
	# group (so a PR still supersedes its own earlier run); push, dispatch and
	# schedule each get their own lane.
	#
	# THE BRANCH RESTRICTION IS GONE, because it contradicted the sentence above.
	#
	# The suffix used to apply only when `ref_name` was `main` or `development`,
	# so on every OTHER branch push and pull_request computed the SAME group —
	# and `cancel-in-progress` made them kill each other. That became reachable
	# when the push allow-list widened on 2026-08-14 to include `feat/**`,
	# `fix/`, `perf/`, `refactor/` and `chore/`: those branches now get
	# both a push run and a pull_request run for one commit.
	#
	# `quality / Quality Report` is a `needs:`-gated aggregator and reports
	# FAILURE when its dependencies are CANCELLED, so the collision showed up as
	# a red gate on a PR that had never actually been evaluated — and re-running
	# collided the same way. See .github#563.
	#
	# A branch name is not a unique lane when two event types can each produce a
	# run for it, so the event is now always part of the key.
	group: quality-${{ github.head_ref \|\| github.ref_name }}${{ github.event_name != 'pull_request' && format('-{0}', github.event_name) \|\| '' }}

	# PUSH RUNS ARE NOT CANCELLED — and this has to be said HERE, not only in the
	# shared workflow. .github#597 set `cancel-in-progress` on quality.yml itself,
	# but a caller's own concurrency cancels the whole run before the called
	# workflow's setting can apply, so that fix reached only the apps that declare
	# no concurrency of their own. Measured 2026-08-28 over push runs on
	# `development` since #597: 0 of 11 cancelled where the caller was silent, 7 of
	# 13 (54%) cancelled where the caller still said `true`.
	#
	# An integration branch needs a verdict per commit: the run being cancelled is
	# the only thing that would have said whether what just landed is sound, and
	# its replacement is cancelled too. `pull_request` keeps cancelling, where
	# superseding really is correct.
	cancel-in-progress: ${{ github.event_name != 'push' }}

	# Permission CEILING for the called quality pipeline. GitHub statically
	# validates the called workflow's declared job permissions against this
	# grant — even for jobs that are disabled — so capping any of these at
	# `read` makes the whole call fail to START (zero jobs, no annotations).
	# Union of what the nested jobs need: journeydoc-capture (contents+actions
	# write), update-baseline / features-extract (contents write), the Quality
	# Report PR comment (issues / pull-requests write), and packages: read for
	# pulling org images.
	permissions:
	contents: write
	actions: write
	issues: write
	pull-requests: write
	packages: read

	jobs:
	quality:
	# Skip the duplicate run a branch-creation push fires alongside the PR event.
	if: github.event_name != 'push' \|\| github.event.created != true
	uses: ConductionNL/.github/.github/workflows/quality.yml@main
	with:
	app-name: integriq
	# composer.json `config.platform.php` is pinned to 8.3, so phpcs/phpmd/etc
	# need the matching runtime or `vendor/composer/platform_check.php` aborts
	# with `Composer dependencies require a PHP version ">= 8.3.0"`.
	php-version: "8.3"
	# Set explicitly, because the shared workflow defaults to
	# `["stable31", "stable32"]` and stable31 CANNOT WORK here. integriq
	# declares <app>openregister</app> as a hard dependency, and openregister
	# declares min-version="32". On NC31 the E2E job logs
	# App "Open Register" cannot be installed because it is not compatible
	# with this version of the server.
	# then continues with a WARNING, so the run proceeds without its central
	# dependency and fails ~70s later on missing tables — which reads like a
	# migration fault rather than a missing app. Testing a version the app
	# cannot support produces red that says nothing. See issue #1172.
	#
	# WHY THIS IS NOW stable34 AND NOT `["stable32", "stable33"]`. This PR
	# moves the dev requirement to `nextcloud/ocp ^34`, so the stubs every
	# static-analysis leg resolves against are NC34's. Running the suite
	# against an older server would check the app under one set of API
	# signatures and analyse it under another, and the two disagreeing is
	# exactly the class of defect the analysers exist to find.
	#
	# ✅ THE NARROWING IS CLOSED, THE FIRST WAY. The previous revision of this
	# comment recorded a deliberate trade: info.xml declared three server
	# versions while the suite exercised one, leaving NC32 and NC33 unmeasured
	# rather than known-broken. It named its own closure condition — "restoring
	# the older refs alongside stable34 once ocp ^34 stubs are proven compatible
	# with them" — and that condition is now met: portaliq runs stable32,
	# stable33 and stable34 against `nextcloud/ocp ^34` and all six PHPUnit
	# legs pass (run 31599055849). So the refs come back rather than info.xml's
	# floor going up, and the two numbers agree again.
	nextcloud-test-refs: '["stable34", "stable32", "stable33"]'
	# `hydra-gates-require-full-coverage: false` is REMOVED here, exactly as
	# the comment it replaced instructed: "Remove this line in the change
	# that wires those producers up, so the coverage requirement arrives with
	# the coverage."
	#
	# WHAT THE FLAG ACTUALLY DID. It is not a gate and it does not decide any
	# finding. It answers one question: when a gate DID NOT RUN, is that a
	# failure? The gate package sorts every non-run into three categories —
	# `na` (subject matter absent: nothing is unverified), `structural`
	# (subject matter EXISTS, nothing produced the gate's input) and `wiring`
	# (the gate's own helper or tool is missing). Only `structural` and
	# `wiring` count against coverage. With the flag set to false, a gate
	# whose machinery had quietly stopped existing reported as green.
	#
	# That is not hypothetical in this fleet: a missing helper made gate-7
	# report PASS over 11 real unguarded IDOR endpoints, and openregister
	# reported `success` while its own log read "EXCEPT GATES 4 33, WHICH DID
	# NOT RUN". A waiver here means integriq's green does not include
	# the coverage assertion at all.
	#
	# THE ORIGINAL DEFERRAL NO LONGER HOLDS, and not for the reason this PR
	# first gave. Re-measured on `development` @5ab1c9df, FULL TREE, with the
	# gate package cloned fresh from ConductionNL/.github@main (756fe89) —
	# not the local submodule, which was 38 commits behind and is therefore a
	# different program:
	#
	# [hydra-gates] COVERAGE: 63 of 64 declared gates reported a result
	# (1 not applicable to this repo/diff; 63 of 63 applicable gates ran).
	#
	# ZERO structural skips, ZERO wiring skips. The single not-applicable was
	# gate-33 axe-core, which declared itself so because this caller had not
	# set `enable-axe`.
	#
	# THAT IS NO LONGER TRUE: `enable-axe: true` is set below, and gate-33
	# now reports a real verdict instead of `na`. CONFIRMED IN CI on the PR
	# that turned it on (#1211, job 93785212530):
	#
	# [gate-33] axe-core: PASS
	# [hydra-gates] gate-33 axe-core: report read — 0 violation(s)
	# present, 0 serious/critical. A PASS here is a PASS
	# over that number, not over silence.
	#
	# "report read" is the part that matters: the report survived the
	# provenance check rather than being rejected and deleted, which is the
	# failure mode where gate-33 goes back to skipping loudly.
	#
	# DO NOT READ THE COVERAGE LINE AS A CONSTANT. An earlier draft of this
	# comment predicted it would move from `63 of 64` to `64 of 64`. That
	# was wrong, and wrong in the way this fleet keeps getting caught by:
	# the count is a function of the DIFF, not of the repo. The same run
	# that produced the PASS above printed
	#
	# [hydra-gates] COVERAGE: 28 of 64 declared gates reported a result
	# (36 not applicable to this repo/diff; 28 of 28 applicable gates ran).
	#
	# because a two-file PR gives 36 gates no subject matter. The `63 of 64`
	# figure quoted higher up came from a FULL-SCOPE run. Compare like with
	# like or the number means nothing. What is genuinely comparable is that
	# gate-33 is now inside the applicable set in both scopes instead of
	# declaring itself `na`.
	#
	# ⚠️ AND THE SAME CAVEAT USED TO APPLY TO A GATE THAT PRINTS `PASS`.
	# gate-16 spec-coverage takes its base from `HYDRA_GATE_BASE_REF`, and
	# the runner supplied none on `--full`, so a full-scope run ON
	# `development` diffed the branch against itself, inspected nothing, and
	# printed PASS — a PASS that COUNTED toward "N of N applicable gates
	# ran" (ConductionNL/.github#361). #364 has since MERGED: a
	# full-repo run with no diff now reports
	# `NOT APPLICABLE — full-repo run computed NO diff … This is NOT a
	# pass.` So do not read a historical green gate-16 cell on this repo as
	# a measurement — re-measure with an explicit base.
	#
	# Measured here on `development` @7c1d9c6d with the canonical package
	# @81c8c97 (which still carried the old behaviour), changing ONLY the
	# base:
	#
	# base origin/development (= what CI's --full run uses) -> count=0
	# base origin/beta -> count=0
	# base origin/main -> count=30
	# --mode report (whole tree, no diff at all) -> 30
	#
	# A committed, untagged public method planted in lib/Service and a
	# matching Vue method moved every one of those numbers by exactly +2 and
	# named both plants, so the zeros above are real zeros and not a dead
	# gate. integriq's genuine gate-16 debt is 30 methods, all backend.
	#
	# Gate 4 was never starved by a producer this repo declines to switch
	# on; it declares itself not applicable. And gate-24
	# integration-parity — named by the first version of this comment as the
	# real gap, which it then was — now PASSES: `scripts/check-integration-
	# parity.sh` landed on `development` after this PR was opened, so the
	# parity script arrived exactly as that comment said it should, ahead of
	# the flag being dropped.
	#
	# PROOF THIS SETTING CAN STILL FAIL, rather than being inert. On the same
	# tree, deleting `scripts/check-integration-parity.sh`:
	#
	# [gate-24] integration-parity: SKIPPED (structural) — ... this repo
	# DOES register integration leaves ... server↔JS leaf parity is
	# UNVERIFIED
	# [hydra-gates] COVERAGE: 62 of 63 applicable gates ran
	#
	# which is precisely the shape the runner exits 98 on, and which
	# quality.yml turns into `::error::hydra-gates passed every gate that
	# ran, but a gate whose SUBJECT MATTER EXISTS did not report`. So this
	# repo has live subject matter for the assertion; it is not being turned
	# on over nothing.
	#
	# NOTE ON WHAT THIS DOES NOT CHANGE. Exit 98 is reached only when no
	# gate FAILED — a run with failures is already red and exits with the
	# failure count. Removing this line therefore adds a verdict where there
	# was none; it cannot mask one. The same full-tree run still shows
	# gate-38 and gate-57 failing on inherited debt in files no PR is
	# touching; both are tracked separately and neither is in scope for a
	# diff-scoped CI run.
	enable-phpcs: true
	enable-psalm: true
	enable-phpstan: true
	enable-phpmetrics: true
	enable-frontend: true
	enable-eslint: true

	# Build the bundle in CI. Nothing here did until now, and the gap hid two
	# real defects for weeks: @nextcloud/dialogs pinned at a vue@2.7 release
	# inside a Vue 3 app, and fifty NcSelect/NcTextArea listeners still on the
	# v8 `@input` contract that v9 does not emit. Both compile, lint and pass
	# unit tests — only an actual build (and, for the listeners, using the
	# control) shows them.
	#
	# A broken bundle is also silent in production: Nextcloud serves whatever
	# js/ was last committed, so the app keeps working from a stale build
	# while main cannot be rebuilt at all.
	# `check:specs` and `test:l10n` are added to this list below rather than
	# here; see the "Frontend Check legs" block further down. The value lives
	# in exactly one place — a second `frontend-checks:` key in the same
	# `with:` block would be accepted last-one-wins by every YAML parser
	# involved and would read as configured while being decided elsewhere.
	# Integration + E2E run against a live Nextcloud server with OR
	# checked out as an additional app (collections + journeys exercise
	# /index.php/apps/openregister/api/objects/integriq/*).
	enable-newman: true
	newman-collection-path: "tests/postman"
	newman-environment-path: "tests/postman/integriq.postman_environment.json"
	# WHY THIS SEED EXISTS. The collection's "01 — Fixture setup (via OR)"
	# folder POSTs to /apps/openregister/api/objects/integriq/<schema>;
	# since the chain-C cutover every integriq entity is an OpenRegister
	# object. Without a seed step the `openconnector` register does not
	# exist, all 8 fixture POSTs return 404, `{{fxSourceUuid}}` is never set,
	# and every downstream request then 404s on the literal placeholder —
	# 51 of 107 assertions (measured, run 30816264169). SEED_SCOPE=register
	# reuses the Playwright job's tests/e2e/ci-seed.sh and stops after the
	# register import, skipping the SPA warm-up and bundle gate that this
	# job (which never runs `npm run build`) must not execute.
	#
	# This was held back until ConductionNL/.github#132 landed: the shared
	# workflow used to start this job's `php -S` SINGLE-WORKER, and journey
	# J2 points a Source at this very instance, so the app called back into
	# the server while the one worker was still inside the outer request —
	# a deadlock that hung the job until timeout-minutes (run 30821823343).
	# #132 sets PHP_CLI_SERVER_WORKERS=8 on this job's server, which is what
	# makes the self-referential fixtures serveable. Merged 2026-08-03.
	newman-seed-command: 'SEED_SCOPE=register bash apps/integriq/tests/e2e/ci-seed.sh'
	additional-apps: '[{"repo":"ConductionNL/openregister","app":"openregister","ref":"development"}]'

	# ── E2E browser tests ────────────────────────────────────────────────
	# `playwright-test-path` does double duty in the shared workflow:
	# 1. it is the directory the "Validate Playwright tests exist" step
	# counts *.spec.ts in;
	# 2. it is the FIRST place the run step looks for a config —
	# `${playwright-test-path}/playwright.config.ts`, falling back to
	# the repo root only if that file is absent.
	# It used to say `tests/e2e/regression`, which holds no config, so lookup
	# (2) fell through to the ROOT config. The run step passes no
	# `--project`, so all four root projects ran at once — 243 tests in 1.6h,
	# including `visual` (pixel baselines the root config itself documents as
	# un-matchable on a CI Linux runner) and `docs-capture` (screenshot
	# re-shoots that have their own dedicated job). `tests/e2e` makes lookup
	# (1) hit tests/e2e/playwright.config.ts, which declares exactly one
	# project over spec-coverage/ + regression/ + workflows/.
	enable-playwright: true
	playwright-test-path: tests/e2e
	# integriq's sources/mappings/synchronizations/jobs are OpenRegister
	# OBJECTS — there is no `oc_openconnector_*` table for them — so with no
	# `openconnector` register the SPA has nothing to resolve and the suite
	# reports it as a wall of selector timeouts. `occ app:enable` is not a
	# reliable provisioning path for it: the InitializeRegister repair step
	# swallows its own failures as warnings and occ still exits 0. The script
	# provisions explicitly and fails loudly when the register or its schemas
	# are still absent. cwd for this step is the Nextcloud server root.
	playwright-seed-command: 'bash apps/integriq/tests/e2e/ci-seed.sh'

	# ── PHPUnit ──────────────────────────────────────────────────────────
	# `enable-phpunit` defaults to FALSE in the shared workflow and this
	# caller never set it, so the job reported `skipped` — with its matrix
	# placeholders still literal (`PHPUnit (PHP ${{ matrix.php-version }},
	# NC ${{ matrix.nextcloud-ref }})`), because GitHub evaluates a matrix
	# job's `if:` before expanding the matrix. A disabled PHPUnit therefore
	# looks exactly like a repo with no PHP tests. This repo has phpunit.xml
	# and a tests/ tree, and a `coverage-report` artifact from 2026-08-03
	# proves the suite does run: 20141 of 33162 statements covered.
	# This is also the prerequisite for the coverage ratchet — the
	# "Coverage Baseline Check" job `needs: phpunit` and downloads its
	# artifact, so enabling the guard without this would be a dead gate of a
	# different shape.
	enable-phpunit: true

	# ── Frontend Check legs ──────────────────────────────────────────────
	# `frontend-checks` was already non-empty (`["build"]`), so the job did
	# run here — but only the build. The validator family ran nowhere.
	# `check:specs` is the aggregate (json-strict + manifest + register) and
	# is listed as ONE leg rather than three, because each leg is a fresh job
	# with its own checkout + `npm ci`.
	# Measured on this tree before enabling: `check:specs` PASSES and
	# `test:l10n` PASSES.
	# `test:l10n:parity` is deliberately NOT added: measured on this tree it
	# is short 337+ translations across the required locales. That is a
	# translation backlog, and a permanently-red leg is one that gets
	# switched off again.
	#
	# `format` (prettier --check) is listed because the shared workflow has NO
	# prettier job of its own — `quality.yml` runs eslint and stylelint and
	# mentions prettier ZERO times. This repo already carries
	# `@nextcloud/prettier-config` and a `format` script, so without this leg
	# `npm run format` never runs outside a developer's editor and the tree
	# drifts straight back out of format between merges — the same inert-
	# formatter failure mode that made the old `.prettierrc` worth deleting.
	# Centralising the config never stopped drift; the gate does.
	# Measured on this tree before enabling: PASSES, 176 of 222 tracked
	# frontend files in scope (l10n/ and docs/ excluded via .prettierignore /
	# .gitignore, which prettier 3 also reads).
	# `check:schema-l10n` is a RATCHET, not a gate. Every string inside a form
	# comes from the schema and is a key in THIS app's catalogue; an absent key
	# renders the English source inside an otherwise translated form, silently.
	# The fleet had 30,459 such strings, so this records the current count and
	# fails only when it GROWS — burning it down stays an ordinary PR.
	frontend-checks: '["build", "check:specs", "test:l10n", "format", "check:schema-l10n", "check:l10n-js"]'

	# ── Hydra mechanical gates ───────────────────────────────────────────
	# `enable-hydra-gates` defaults to FALSE, so this tier has never executed
	# here — the job reported `skipped`, which the Quality Report renders
	# identically to a pass.
	enable-hydra-gates: true
	# No `hydra-gates-ref` here on purpose. The shared workflow defaults it
	# to @main, and this workflow is itself consumed at @main, so the two
	# sides move together and a gate fix reaches this repo without a commit
	# in this repo. A pin is a silent expiry date: 22 repos sat on v1.0.1 and
	# 16 gates were dead fleet-wide while every one reported PASS (.github#159),
	# and a default flipped at @main later reached those old runners and made
	# them red on gates they had no subject matter for (.github#173).
	# To hold this repo still for a specific reason, set the input explicitly
	# and say why — it is still honoured. To roll back for everyone, revert on
	# ConductionNL/.github main.
	#
	# THIRD CAUSE, and the one that is failing this repo RIGHT NOW
	# (.github#177): quality.yml@main began executing three gate helpers BY
	# NAME — check_spec_anchors.py, check_form_labels.py and
	# check_license_triangle.py — which exist in NO tag before v1.5.0.
	# Verified by DIRECTORY LISTING of each tag, not by per-file lookups:
	# those answered "present" uniformly across v1.0.0..v1.5.0, and the
	# uniformity across independent inputs was the tell that the instrument
	# was wrong. So the Hydra Gates job here fails at "Verify the pinned gates
	# package satisfies this workflow", before a single gate runs, with the
	# workflow's own words: "This is NOT a code-quality finding about your
	# repository." Removing the pin is the repair.
	#
	# Unpinning also picks up v1.5.1's push scoping (.github#179): on a push to
	# `development`, `origin/development` IS `HEAD`, so the diff was empty by
	# construction — <= v1.4.0 passed over it (permanently green) and v1.5.0
	# refused with exit 99 (permanently red). The scope is now
	# `github.event.before...HEAD`, what the push actually changed.
	# ── axe-core (gate-33) ───────────────────────────────────────────────
	# `enable-axe` used to be deliberately unset here, on the grounds that a
	# vanilla Nextcloud 34 reports serious/critical violations on core's OWN
	# routes that DOM scoping does not remove. That reasoning is now out of
	# date in the part that mattered: `axe-include-selector` defaults to
	# `#content-vue, #content`, which scopes the analysis to the app's own
	# rendered DOM, and the runner proves on every run that the scope is a
	# SCOPE and not a MUTE (it injects a violation inside the container and
	# asserts it is reported, and one outside and asserts it is not).
	#
	# So this is turned on against a MEASUREMENT, not a hope. Run locally
	# with the canonical runner (ConductionNL/.github@main b8c7ead,
	# @axe-core/playwright 4.12.1) against a dedicated Nextcloud 34 with
	# openregister + integriq installed and the register seeded:
	#
	# /index.php/apps/integriq/ HTTP 200 passes=19 violations=0
	# /index.php/apps/integriq/#/sources HTTP 200 passes=17 violations=0
	#
	# with `axe self-test OK` and `axe scope control OK` on both. Zero
	# violations at any impact, so zero serious/critical — gate-33 has
	# nothing to fail on today.
	enable-axe: true
	# ONE ROUTE, AND THAT IS A RUNNER DEFECT, NOT A CHOICE — see
	# ConductionNL/.github#351.
	#
	# `axe-run.cjs` treats a null `page.goto()` response as HTTP 0 and dies.
	# A fragment-only navigation returns null because no document is fetched,
	# so for a hash-routed SPA — which this app is (`createWebHashHistory()`)
	# — every route after the first one shares a document with it and is
	# reported as `returned HTTP 0. … Set the axe-routes input to routes this
	# app actually serves.` The route is served fine; only its POSITION in
	# the list decides. Proven both ways on the same instance: `#/sources`
	# listed FIRST analysed cleanly at HTTP 200; the SAME route listed after
	# the app root killed the run with exit 2 and wrote no report at all.
	#
	# That is the trap in this input: the error blames the caller, so the
	# obvious response is to delete routes until it goes green, which
	# silently narrows accessibility coverage to the landing page and reads
	# as a config fix. Recording it here so the next person widens the list
	# by fixing #351 rather than by guessing.
	#
	# Until #351 lands this is the app root only. It is a real verdict on a
	# real page — gate-33 has reported SKIPPED in this repo since it was
	# written — but it is one page, not the app.
	axe-routes: "/index.php/apps/integriq/"

	# ── Coverage ratchet ─────────────────────────────────────────────────
	# `enable-coverage-guard` defaults to FALSE in the shared workflow, so
	# BOTH coverage jobs — "Coverage Baseline Protection" (PR) and "Coverage
	# Baseline Check" (push) — have only ever reported `skipped` here. A
	# skipped job renders in the Quality Report exactly like a passing one,
	# so the ratchet read as present while checking nothing.
	#
	# Turning it on needs two files this repo did not have, both now added
	# alongside this line:
	# * `scripts/coverage-guard.php` — copied BYTE-IDENTICAL from
	# openregister (the same copy softwarecatalog took), so there is one
	# implementation of the ratchet across the fleet rather than five.
	# * `.coverage-baseline` — a bare number. It must EQUAL what CI
	# measures on development, not approximate it: too high and the
	# "Guard coverage baseline" step inside PHPUnit exits 1; too low and
	# the push-side Coverage Baseline Check recomputes a higher value,
	# finds `git diff` non-empty, and fails demanding the new number be
	# committed. There is no safe margin in either direction.
	# The committed value is the one CI printed, recorded as measured.
	#
	# This was held back until `enable-phpunit` above had produced a
	# `coverage-report` artifact from this repo, since the guard reads the
	# project-level metrics out of that job's clover.xml.
	enable-coverage-guard: true

Navigation Menu

Uh oh!

chore(release): 0.3.15-unstable.20260831165924 #2369

Workflow file

chore(release): 0.3.15-unstable.20260831165924 #2369

Uh oh!

Workflow file for this run