From 74d1df23ff206482ceb052bf5be03e6740325418 Mon Sep 17 00:00:00 2001 From: Conduction Release Bot Date: Thu, 10 Sep 2026 11:52:52 +0200 Subject: [PATCH] feat(quality): let a repo opt into gate-46 blocking on @e2e anchors MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Gate-46 resolves `@spec` and `@e2e` by identical rules, but only `@spec` blocks. The `@e2e` half is report-only, and its warning ends "set HYDRA_GATE_SPEC_ANCHOR_E2E_BLOCKING=1 for this repo once they are worked down". That instruction was unreachable. The gate step's `env:` block is fixed in quality.yml, and a reusable workflow does not inherit `env` from its caller, so no app repo could set the variable however it wrote its own workflow. Measured across the organisation today: the string appears nowhere except the runner and the runner's own test. Zero repos had opted in, and none could have — the gate has been inviting an opt-in that did not exist. This adds the input that accepts it. Default false, and it must stay false: #726 counted 194 dangling @e2e anchors fleet-wide with 154 of them in four repos, which are the four with the most end-to-end coverage. Blocking on day one would turn them red on debt no PR introduced. The value is passed as an explicit '1'/'0' rather than the boolean's own string form. The runner compares against the literal "1"; a GitHub boolean interpolates as "true"/"false", which reads as "not 1" and would leave the gate report-only in a repo that had opted in — a flag that says it is on, a gate that is off, and nothing in the log to tell them apart. One input per flag, deliberately. A single passthrough for arbitrary HYDRA_GATE_* variables would also let a caller silently switch OFF a blocking gate, which is the opposite of what these opt-ins are for. Gates 112, 113, 114 and 115 carry the identical invitation and are equally unreachable today. They are not wired here — each is a separate decision about a separate backlog, and this one has a repo behind it: openregister's @e2e count is measured at zero. Runner side unchanged and already proven: scripts/lib/test_gate_46_tests_scope.sh asserts "HYDRA_GATE_SPEC_ANCHOR_E2E_BLOCKING=1 turns the warning into a failure", and the suite is green on this branch. --- .github/workflows/quality.yml | 35 +++++++++++++++++++++++++++++++++++ 1 file changed, 35 insertions(+) diff --git a/.github/workflows/quality.yml b/.github/workflows/quality.yml index ffeefb8f3..e0e7a955d 100644 --- a/.github/workflows/quality.yml +++ b/.github/workflows/quality.yml @@ -557,6 +557,30 @@ on: required: false type: boolean default: true + hydra-gates-e2e-anchor-blocking: + # THE OPT-IN GATE-46 ALREADY INVITES AND NOTHING COULD ACCEPT. + # + # Gate-46 resolves both traceability tags by identical rules, but only + # `@spec` blocks; `@e2e` is report-only, and its warning ends "set + # HYDRA_GATE_SPEC_ANCHOR_E2E_BLOCKING=1 for this repo once they are + # worked down". That instruction was unreachable. The gate step's `env:` + # block is fixed in this file, and a reusable workflow does not inherit + # `env` from its caller, so no app repo could set the variable however + # it wrote its own workflow. Measured 2026-09-10: the string appears + # nowhere in the organisation except the runner and the runner's own + # test — zero repos had opted in, and none could have. + # + # Same shape and the same reasoning as gates 112, 113, 114 and 115, + # which carry the identical "set HYDRA_GATE__BLOCKING=1 for this + # repo" invitation and are, today, equally unreachable. Wiring one flag + # per input is deliberate: a single passthrough for arbitrary + # HYDRA_GATE_* variables would let a caller silently disable a blocking + # gate, which is the opposite of what these opt-ins are for. Each new + # one is a visible, named line in the caller's workflow. + description: "Make gate-46 (spec-anchor-existence) BLOCK on unresolved `@e2e` anchors, not just on `@spec`. OPT-IN, default false, and it must stay opt-in: #726 counted 194 dangling @e2e anchors fleet-wide with 154 of them in four repos (dossiq 46, zaakafhandelapp 43, pipelinq 36, decidesk 29), so blocking on day one would turn the four repos with the MOST end-to-end coverage red on debt no PR introduced. The count is printed on every run either way — turning this on changes the verdict, not the visibility. Turn it on for a repo only once its @e2e count is measured at zero, with the gate's own resolver (scripts/lib/check_spec_anchors.py) rather than a reimplementation: three published figures for a comparable set were wrong on 2026-09-10 and every wrong one came from a hand-rolled resolver. `@spec` blocks regardless of this flag, exactly as it has since #246." + required: false + type: boolean + default: false # ── Cost controls (measured 2026-08-27, .github#5xx) ────────────────── # @@ -7296,6 +7320,17 @@ jobs: # The author's own head commit, so the fallback stops reading the # merge commit. Empty on a push, where `git log -1` is already right. HYDRA_GATE_HEAD_SHA: ${{ github.event.pull_request.head.sha }} + # Gate-46's `@e2e` half blocks only where the caller asked for it. + # + # Passed as an explicit '1'/'0' rather than the boolean's own string + # form. The runner compares against the literal "1"; a GitHub boolean + # interpolates as "true"/"false", which would read as "not 1" and + # leave the gate report-only in a repo that had opted in — a flag that + # says it is on, a gate that is off, and nothing in the log to tell + # them apart. Set unconditionally so the variable's value always comes + # from the caller's declaration rather than from whatever the runner + # defaults to. + HYDRA_GATE_SPEC_ANCHOR_E2E_BLOCKING: ${{ inputs.hydra-gates-e2e-anchor-blocking && '1' || '0' }} run: | set -u ARGS="--app-dir ${GITHUB_WORKSPACE}/app"