rastrillo / fiesta Public

Test evidence for Task 11 of the Fiesta packaging, copied from the implementer's report so it sits on the branch.

Commits

• abb0409 Retarget the shell drives at the layout examples

Preservation: every ported test passed on its first run

One run, in-tree, after the port and before any control, over the 35 tests:

ok  	amadan.net/rastrillo/fiesta/check	70.125s

35 top-level --- PASS, 0 --- FAIL; nothing loosened, nothing re-run. Heaviest: TestTheProfileMenuIsADropUpAtTheRailsFoot 16.60s, TestThePhoneIndexWorksWithScripts 7.73s. The prefetch leg arose: the index was fetched with a prerender Sec-Purpose and delivered as navigational-prefetch.

The slot test, and its control

TestEveryLayoutHasOneDefaultAndOneSlot already covers every layout-*; in-tree --- PASS (0.00s). Control, in a scratch copy, both slot comments removed from examples/layout-topbar/default.html:

    example_test.go:570: layout-topbar/default: example: the document holds 0 slots, want one
--- FAIL: TestEveryLayoutHasOneDefaultAndOneSlot (0.00s)
    --- PASS: TestEveryLayoutHasOneDefaultAndOneSlot/control (0.00s)

The rail control

The brief's mutation has nothing to act on: the rail has no inline-size declaration (its width is the sidebar grid's first column, grid-template-columns: 15rem 1fr), and shell_browser_test.go has no rail-width assertion. Run as the grid column set to auto in a scratch css/tokens.css, over all 35 ported drives:

--- FAIL: TestTheWideLayoutIsUnchanged (2.74s)
    --- FAIL: TestTheWideLayoutIsUnchanged/sidebar_ltr (0.63s)
    --- FAIL: TestTheWideLayoutIsUnchanged/sidebar_rtl (0.79s)
        index_browser_test.go:254: sidebar ltr /index at 1280: rail 0,0,102,900 main 102,0,1163,900; the old layout gives rail 0,0,121,900 main 121,0,1144,900
--- FAIL: TestTheProfileMenuIsADropUpAtTheRailsFoot (54.86s)
    (12 of 24 legs: the ltr ones)
        profile_browser_test.go:307: open: card shown true, its bottom at 664 over an avatar whose top is 668, on screen false; want a drop-up wholly in the window

The variant shell_browser_test.go's own rail drive sees, the rail's box-sizing back to content-box:

    shell_browser_test.go:356: the rail's border box is 932px in a 900px viewport: it is 32px taller than the window it is sized against, so whatever sits at its foot is under the fold. box-sizing on the rail is the fix
    shell_browser_test.go:360: the sticky rail's bottom edge is 32px below the foot of a 900px window
    shell_browser_test.go:365: the person block at the rail's foot ends 16px below the foot of the window: it is off-screen with nothing to scroll to reach it
    shell_browser_test.go:459: the rail's border box is 452px in a 420px viewport: a short window overflows for the same reason a tall one did
    shell_browser_test.go:484: the person block is still 16px below the window after the rail was scrolled to its bottom: it is unreachable, not merely off-screen
--- FAIL: TestTheSidebarRailPutsThePersonAtItsFootAndTheProfileMenuOpensUpward (0.67s)

Shards and the gate

go vet -tags browser ./check/ and make staticcheck clean; the untagged TestParityLinksAreRewritten (with the new dir = "/" case) and TestEveryFixtureRowResolves green.

make browser.check~1of3   170s   91 of 273 tests   ok
make browser.check~2of3   158s   ok
make browser.check~3of3   162s   ok
make ci                   exit 0, 494s (check shards 160/154/166s)