Skip to content

test(qcm): mixed-provider camera scene with one layer failing (QQA-10b) - #45

Open
joepetjr wants to merge 1 commit into
qqa-16c-camera-polishfrom
qqa-10b-mixed-scene
Open

joepetjr wants to merge 1 commit into
qqa-16c-camera-polishfrom
qqa-10b-mixed-scene

Conversation

@joepetjr

Copy link
Copy Markdown
Contributor

Ported from headlessmode-archive #806 (commit 17b12c5b), QC U13 camera chain (QC-04), plan row U13. Base qqa-16c-camera-polish (PR for #801), stacked. Closed in the archive, not merged there; this is the port.

What changes: test only. A real-browser proof that road, park, camera-link, research and airport layers draw together at one point, and that one failing layer leaves the others working. No source change; the helper change (camera-snapshot-stub.mjs) is additive. Because it is test-only there is no red run.

Adapted to the shipped catalogue: the stand-in server also answers master's place-name, road and water queries (/(geonames|transportation|nhd)/MapServer/N/query) with an empty feature list so those default layers settle; site-coupling.json is regenerated.

Tests (through bin/pytest-slot.sh, logs unpiped):

  • Green: the new browser test plus register, boundary and docs checks, 39 of 39 pass.
  • CI's node step, full (built site, browser tests ran), on this branch before opening: 3141 tests, 3141 pass, 0 fail, 0 skipped, exit 0. Python subset (front matter, content validation, privacy accuracy, headers wiring, a11y config, data contract, public tree): py_rc=0. The full run was with master at fdfc050; master then took fix(qcm): put the last seven map reads on fetchWithDeadline #38 (adapter reads on fetchWithDeadline), so this branch was rebased onto 530a556 (clean) and the boundary, no-raw-fetch, adapter-deadlines, control-register, acceptance-manifest and docs-promises checks re-run on the final head: all pass (log quick2.log).

Puts the road, park, camera-link, research and airport layers on at one
public point and checks them together in a real browser. Each layer draws
its own pin class, the count line reads five layers on, and the Cameras
list names each row's own source. Opening one pin per provider applies that
provider's own picture plan: a blob image for the road camera, a plain
image pointed at the owner for park, research and airport, and no image and
no request to the owner for the link-only layer. Each card names its own
operator.

Then one layer's file fails while the others load. That layer's row says
the list is unavailable, the other layers keep their pins and pictures, the
list says it is partial, and a surviving row still opens its card. Cycling
the layer off and on brings it back. The road layer failing alone, and all
five failing at once (unavailable, never partial), are covered as well.

The fixture files reuse each provider's real header from the committed
seed with made-up camera rows around the point; the helper change is
additive. No source change was needed: the run found no defect.

Ported from headlessmode-archive #806 (commit 17b12c5b), stacked on the QQA-16c camera list PR.
@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying headlessmode-current with  Cloudflare Pages  Cloudflare Pages

Latest commit: e20342b
Status: ✅  Deploy successful!
Preview URL: https://26a84f4e.headlessmode-current.pages.dev
Branch Preview URL: https://qqa-10b-mixed-scene.headlessmode-current.pages.dev

View logs

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant