Repository navigation
test(e2e): cover the verify skill's undriven flows and run the suite in the merge queue - #217
Merged
Merged
Conversation
Upd4ting
marked this pull request as ready for review
November 19, 2026 04:51
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Wdf26acfc-19. Every browser test of dms-database belongs in its Playwright suite (
playwright test -c e2e). Pull requests run the@smokeset, and the merge queue runs every journey. Two gaps remained. The verify skill still listed main user flows as "Not driven". And the checks dms-control ran outside a browser had stopped running once the journeys left the skill:dms-control mapand the npm pack, locale and range checks.What changed
CI (
ci.yml).selectoutputse2e(none,smokeorall) andshards:--grep @smoke, 1 runnerci:full,merge_group,workflow_dispatch, push, nightlye2e-reportnextThe
ubuntugate requirese2ewheneverselectasks for it. Theon:triggers belong toantelopejs-sync: the move to@antelopejs/tooling-configs0.8.9 bringsmerge_group, thelabeledtype and the push tomain, and dropsnext-sync.ymlandstale.yml. Before, pull requests tomainand manual runs ran@smoke, the nightly ran every journey, and a pull request intonextran nothing.Stack (
e2e/stack.ts,e2e/ctx.ts). A journey can stop and restart MongoDB (ctx.stopDatabase,ctx.startDatabase) and write its database directly (ctx.database). A reset restarts a stopped replica set, so a failed journey leaves the next one a running database.6 new journeys for flows the feature map listed as not driven (table below).
5 stale journeys updated. They had failed on
main's nightly since feat(database): promote next to stable #216, because the product and docs changed and the journeys did not. A numbercontainsfilter now compares the whole value (docs/4.usage/4.data-browser.md:16). An external database now gets a Diagram (docs/4.usage/6.external-databases.md:40-42). The search skips list columns. No product code changed.Non-browser checks become mocha tests in
pnpm test.src/test/package.test.tsreplaces the 3 package journeys. It lists the tarball withnpm pack --dry-run --json, so it also runs on Windows, and checks the locales and the README ranges.src/test/feature-map.test.tsdoes whatdms-control mapchecked: every user-facing file maps to a feature, every docs page exists, and every journey is named in a feature file. On its first run it found 12 unmapped files and 1 undocumented journey. Both are fixed in the feature files.The verify skill's SKILL.md, the feature files and AGENTS.md describe the new state.
Proof
Verify feature-map flows and the spec that covers each
Each new journey passed
--repeat-each 3locally, and the Overview one passed 5 of 5.specs/overview.spec.tsan-unreachable-database-shows-unavailable-and-check-again-recoversspecs/library.spec.tshistory-groups-runs-by-day-with-yesterday-and-older-datesspecs/library.spec.tsa-history-row-without-a-status-reads-as-okspecs/editing.spec.tsan-undo-whose-write-fails-keeps-the-saved-values-and-offers-undo-againspecs/diagram.spec.tsescape-leaves-the-find-field-then-lifts-the-focus-then-clears-the-selection/modules/database/query?schema=&table=opens the console on the table, per instancespecs/query.spec.tsa-link-opens-the-console-on-the-table-in-the-instance-it-namesspecs/query.spec.tsctrl-space-inside-a-filter-completes-the-columns-of-the-table,format-strips-trailing-spaces-and-blank-lines,the-chart-plots-the-first-numeric-column-with-pickers-for-both-axes,open-in-data-browser-opens-the-target-table-or-the-row); only SKILL.md was staleStill not driven, with the reason in SKILL.md: the table watch on a standalone MongoDB, the empty schema-cards state, the diagram's name order (vitest covers it), the extension boundary and the role editor.
One known failure, a product finding: with MongoDB down,
GET /api/database/healthanswers 500 after the driver's 30 s server selection, notstatus: downwithin the 3 s probe deadline. The card still turns Unavailable, 60 to 120 s after the stop.Integration tests that replace each non-browser check
the-tarball-holds-what-its-manifest-namessrc/test/package.test.ts, 3 testsboth-locales-hold-the-same-textssrc/test/package.test.tsranges-match-the-compatibility-tablesrc/test/package.test.tsdms-control map: user-facing files covered, docs pages existdms-control map, broken since #198src/test/feature-map.test.ts, 4 testsThe formal checks keep their own runners:
formal-table-watchhas its CI job, andformal-access-write-backis already a test inpnpm test.Playwright test counts
The change is 151 − 3 package checks moved to mocha + 6 new journeys. The
@smokeset is unchanged and fast:smoke,every-route-needs-a-signed-in-session,edits-are-staged-then-reviewed-then-written-one-put-per-row,the-connection-card-reads-the-probeanda-query-returns-its-rows-in-the-order-it-asks.Whole suite
mainfailed in both shards on 17, 18 and 19 Nov, with 5 failing journeys (data ×2, editing ×1, external ×2). A local whole-suite run of this branch before the journey update matched that: 149 passed, 5 failed.@smokee2e job in 2 min 50 s. The new mocha tests ran green inci / Test .nvmrc.Scope
e2e/stack.ts), the pilot that preceded the@antelopejs/dms/e2ekit. It already runsplaywright test -c e2ewithout dms-control. Moving it onto the kit is a follow-up once the kit drops dms-control.Follow-ups
GET /api/database/healthanswers 500 after 30 s becausegetOverviewHealthreads the totals after a failed probe; done means it answersstatus: downwith the probe's error within the 3 s deadline and the journey'sknownFailuresentry is removed.dms_database.data.save.undo_failedis never shown (a failed undo showsundo_partial); done means the key is used for a wholly failed undo or removed from both locales.ci:fulllabel; done means the label exists and labelling a PR runs every journey.e2e/runs its own stack, not the@antelopejs/dms/e2ekit; done means the suite registers its journeys through the kit once the kit no longer needs dms-control, with the same 154 tests.