fix-issue/next-issue had unresolved placeholders (<dein Testkommando>,
<owner>/<repo>) and no explicit branch-creation step, so a run that
skipped the CLAUDE.md convention could commit straight to main. Adds a
CI-gate before setting ai-review (actions_run_read against the
existing .gitea/workflows/ci.yml) instead of trusting self-reported
test results, plus a /work-queue command to chain next-issue -> fix-issue.
dev.bundle is Vaadin-regenerated and was tracked+dirty, so every
autofix commit would have swept in unrelated binary diffs.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0114SVTvBYjLBt3TTxhzJttd
DashboardView and GridStackView rendered the same three charts from the
same hardcoded data, so every chart factory method existed twice. The
dashboard is now the GridStackLayout variant at @Route(""):
- GridStackView deleted, its grid/toolbar/chart factories moved into
DashboardView, which keeps the point-click notifications the old
dashboard had.
- Charts size to their widget (100%/100%) instead of a fixed 400px.
- MainLayout has one dashboard entry; nav.gridstack/page.gridstack keys
dropped from all three bundles.
- Storage key is "dashboard" (was "gridstack-demo"), so a saved layout
from the old route is not reused.
- GridStackViewTest renamed to DashboardViewTest, plus a test asserting
the three charts render.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0124BiJikbhbiEsNfJxdWM69
Adds the first browser-driven test to the project: PlaywrightTestBase starts
the app on a random port and drives a headless Chromium against it, and
FormViewPlaywrightTest covers the form's rendering, binder validation and the
save/reset buttons.
The tests skip themselves via a JUnit assumption when no browser can be
launched, so runners without Playwright browsers stay green.
Closes#16
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ue9ZtWUBQF4SuSHpzZ3zwq
Pit Friedrich