Every text-[color:var(--ink)]/50, border-[color:var(--ink)]/10 and
bg-[color:var(--ink)]/5 in the templates compiled to nothing. Tailwind v3
cannot put an opacity on an arbitrary var() colour, so all 538 of them were
dropped from output.css without a word. Meta text rendered at full ink, card
borders fell back to Tailwind's default grey (bright white rings at night),
and the washes, backdrops and hover pills simply weren't there, on the news
site, the casino and the adventure pages alike.
The palette is now written once per phase and casino room as bare rgb
channels (--ink-rgb and friends), the plain vars are built from them, and
tailwind.config.js exposes ink, muted, accent, warn and card as real colours.
The classes are text-ink/80, border-ink/10, bg-accent/20 now, and they compile.
Restoring the intended opacities as-is would have traded flat for illegible:
ink at 50% measures 2.7 to 2.9:1 on the lit phases. So secondary text gets
its own colour, --muted, per phase and per casino room, taken as faint as it
can go while still clearing 4.5:1 on both the card and the page. Everything
that was ink at 35 to 70% is text-muted. The 20-30% ones are separators and
empty-cell dashes and stay faint.
The bg-theme-* chips carry white text everywhere they appear, and eight of
them measured 2.5 to 3.2:1. Same hues, taken just dark enough to pass (4.6 to
4.7:1). eu, lego and adventure already passed and are unchanged, and the
underline and border colours keep the old pastels since they carry no text.
Static files came out of the embedded FS with no Last-Modified, no ETag and
no Cache-Control, so every page view downloaded the stylesheet and every
script in full, and the service worker's "cheap 304" never happened. Each
file is hashed once now. Templates link through {{asset}}, which appends
?v=<hash>; a request naming the current hash is served immutable for a year,
and everything else gets the hash as its ETag and revalidates to a 304. The
service worker serves versioned URLs cache-first, drops older versions of the
same file as new ones land, and moves to cache v6.
Trivia had every Go test passing and had never been in a browser, which
this plan's own rule says means nothing. So: play it.
The game itself holds up. The clock drains honestly and does not restart
on a reload, the multiple compounds, walking pays exactly what the felt
quoted, the reveal marks the right answer, and the auto-submit at zero
lands as a timeout rather than an illegal move. The next question's
answer never crosses the wire.
Two bugs only the browser could show:
- The spot printed double the stake after every settled game. standing()
set spot.amount and *then* poured the chips on, and pour grows the pile
from what it is told is already there. The money was always right; the
number under the chips was not, which is the one rule the felt is built
on.
- The house rack sat on top of the multiplier at 390px. Its 5.75rem inset
is not a margin, it is the width of blackjack's shoe — so on a phone the
rack sits in the middle of the felt. It now shrinks on small screens and
pulls into the corner where the corner is empty; data-at says which rack
is which, because pulling blackjack's to the edge slides it under the
deck.
The dev rig seeds its own question bank now (one real OpenTDB batch per
difficulty), because a fresh dev database 503s every start otherwise.
prosolis