This is the accrual engine, running in your browser, imported from the same module the API and the tests use. Mint desks, turn the wheel, take payouts — and then switch on the bug that every implementation of this gets wrong the first time and watch the books stop balancing.
reading the service…
None yet. Mint one, then turn the wheel.
Added up from every desk on every slot, against a total this component tracked separately. Nothing here asks the engine whether it is happy.
Mint two desks, turn the wheel three times, then tick the bug on and mint a third. It arrives owed everything the first two earned — and the books go over by exactly what it was handed.
Mint fifty desks and buy a small amount. Each desk's share lands under the floor, every claim is refused, and nothing is lost — the entitlement keeps accruing because it is derived, not stored.
Buy 100 base units across 7 desks. Each gets 14, two units will not divide, and they are carried into the next round rather than given to whoever happens to be first in the array.
It is not reading a chain. Nothing has launched, so there is nothing on-chain to read. Every number here comes from the operations you performed in this tab, and it disappears when you reload.
What it does prove is the arithmetic — that this particular way of paying five thousand desks conserves what it bought, under any interleaving of mints, rounds and claims you can produce by hand. The test suite does the same thing four thousand steps at a time.