[ixuz]

What and why

Since #200, a pull request and the merge queue wait for the slowest journeys job. In the run that tested #200, that was journeys (engines, 2/2) at about 5.8 minutes, and the four jobs took about 20 job-minutes between them. Much of that time checked nothing a faster run did not already check. This follows the issue's proposal, one commit per step:

1. Time the accessibility check apart. Each step's axe check now runs in a step of its own, "Accessibility", so the report times it. Locally it takes 35–67% of every journey's time, Chromium included: more than the issue's estimate.
2. One job per engine. The journeys now run in five jobs side by side: chromium-1 and chromium-2 (Chromium in two shards), firefox, webkit and phone. The slow WebKit and phone journeys no longer land in one job. WebKit's and the phone's browsers share a cache. journeys-report and the warm jobs follow the new names.
3. Axe in Chromium and on the phone only. Desktop Firefox and WebKit leave it out (metadata: { accessibility: false } on their projects). The phone still checks WebKit's rendering, and every screen is checked in Chromium. docs/accessibility.md says what goes unchecked: what only Firefox computes, such as a mixed color's contrast. Locally, Firefox's reload journey went from 31.7 to 17.5 seconds.
4. No smoke journey in the engines. The reload journey now logs out at its end, so each engine still logs in and out once. The engines run 9 journeys instead of 12; Chromium still runs @smoke.
5. The app's logic in Chromium only. App.step(title, body, { in: ["chromium"] }) runs a step only in the projects named, and the report shows it skipped in the others (test.step.skip). It is used for the reload journey's statistics and card history (steps 09–10) and the shared link's import, study and unknown link (steps 04–05). The guest journey still imports and studies in every engine.
6. Chromium runs fully parallel. Each journey is a unit the shards and workers share out, so pod-safety's six journeys, about 100 seconds, are spread over both shards instead of ending one.
7. The Studio journey, split. At 2.4 minutes it was the longest test, and one test cannot be split by a shard or a worker. Writing, checking, test-playing and publishing a course now make a journey of their own, in studio-course.journey.ts, with its own user, instance and studied deck. It is followed, in turn, by the journey that adds the published course from its link. The Studio journey keeps the decks, the workbench, the drafts and the graph. Locally the two take 64 and 28 seconds. docs/user-guide.md and docs/statistics.md point to the new steps.

Every Chromium and Firefox journey passes locally. WebKit and the phone could not run on the machine this was made on (missing system libraries), so this pull request's CI run is the first to run them with these changes, including the reload journey's new logout and the skipped steps. Its timings will show the saving per job.

Not done here: axe also takes about 44% of Chromium's journey time. Checking each screen once instead of at every step, an alternative the issue set aside, may be worth an issue of its own now that it is measured.

Closes: #201

Checklist

• One issue. This pull request addresses one issue, linked above.
• One step. This pull request does not finish its issue, so it says Refs:, not Closes:. Leave this unchecked otherwise.
• Done, as AGENTS.md defines it: npm run check passes (lint, Markdown and spelling included), npm run lint:reuse too if a file was added, and npm run crosscheck too if ns/ or decks/ changed. (npm run lint:workflows passes too; crosscheck n/a: ns/ and decks/ unchanged)
• Tests. New and changed code has its tests, and every package stays at 100% coverage (docs/testing.md). (The changes are to the journeys themselves, which run in Chromium and Firefox locally and in every engine in CI.)
• Docs. The docs say what changed, and a new page in docs/ is listed in docs/docs.ttl. (no new page)
• Languages. A new or changed message is in every locale: en, sv, ko, de, es and fr (packages/ui/src/i18n/). (n/a: no message changed)
• Vocabulary and shapes. Any change to ns/ was agreed in an issue first; no published ns/shapes/*/vN.ttl or decks/*/vN.ttl was edited; generated files were regenerated, not edited. (n/a: nothing in ns/ or decks/ changed)
• Breaking? Nothing in the public API breaks, or a commit says so (! and a BREAKING CHANGE: footer). npm run check:breaking sees a line of api/ that goes, but not a feature removed, nor a reader that reads less than before.
• Pod data. Any change to what the app writes to a pod comes with its migration (docs/migrations.md). (n/a: the app is unchanged)
• Policy. Any change to an external rule the code follows (SM-2, the plural rules, a specification) is cited. (n/a)
• Licenses. New content is MIT, or a deck states its own license; a new file has its SPDX header (npm run lint:reuse), or a new deck its line in REUSE.toml; third-party content is credited.
• Architecture. Imports follow docs/boundaries.md and the layers of docs/architecture.md.
• Run. I ran the app (npm run dev) and tried the change, and considered the user journeys. (The change is to the journeys: every Chromium and Firefox journey was run against the built app; WebKit and the phone run in CI)
• Authorship. I read and understand every line of this pull request, and answer for it, whatever tools helped write it (CONTRIBUTING.md).
• Commits. Each commit is one logical change, in Conventional Commits form (type(scope): subject, ! and a BREAKING CHANGE: footer if it breaks something), with an imperative subject and a body that explains why (CONTRIBUTING.md).