0.8.1.0:0 — bundle Station Master v0.8.1.0, and date the games in Manage Game
A table pass over 0.8.0.17. The second-digit bump is deliberate: 0.8.1 had been reserved for the seatless display table, which is getting more thought, and this batch earned the number on its own. GAMES IN PROGRESS RESUME, AND IT WAS MEASURED BEFORE RELEASE rather than argued from the diff. The bundled release changes what sits in the yards, so the risk was real. All twelve saves on the test box were pulled and replayed through `tryResumeSession` — the server's own boot check — against the new build: six resume, six refuse, and the six refusals are the SAME six, at the same moves, with the same codes, that 0.8.0.17 already logged. Nothing new was stranded. That replay is the check worth repeating on any release that touches state, because it answers before the install rather than after. THE GAME'S OWN FIX, for the README's benefit: the Freight Agent's "clear the red Inbound box" returned the car to the Classification Yard still marked loaded, carrying a load already delivered and already paid for. `pooled()` strips a load's `origin` stamp and keeps `loaded` on purpose — a train can retire at a Division Point with freight aboard — so the call site's comment, "a car back in a yard is back in the common supply, carrying nothing", described something it never did. Coaches suffered worst: detraining takes `coach && !loaded` out of the Division Yard, so a cleared coach came back as stock that could never unload a passenger. 18 loaded coaches against 6 empty over five measured games. THIS PACKAGE'S OWN CHANGE IS THE MANAGE GAME DROPDOWN. It listed each game's code, seat count and position and nothing about when the game was last touched, so four games on a server looked alike — reported 2026-09-22: "I've got four games out there, but I have no idea which one's which or how old they are." `lastMoveAt` was already on the wire from the server's `summary()` and simply unused. Each option now carries both an age and a timestamp, because they answer different questions: "5h ago" answers the one that was asked, and the date beside it is what you quote when deciding to end a game. A lobby has had no move, so it reports when it was created and says so. `when()` moved out of `gamesInProgress.ts` into `utils.ts` alongside a new `ago()`, shared by both actions so an administrator sees one date format wherever a game is named — the second copy would have been the one that drifted. No route, file model or interface changed. README and instructions.md carry the release, and the notes in all five locales say games in progress are unaffected. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MUizFYCMHRWhbWwXhp7WPR
This commit is contained in:
@@ -13,7 +13,30 @@ Station Master is a railroad operations board game with an authoritative multipl
|
||||
solitaire needs no server and is not what this package is for. This package runs that server —
|
||||
the browser client, the lobby, and the intent/SSE API — as a single StartOS service.
|
||||
|
||||
**Bundled version: 0.8.0.17.** Four fixes from a live playtest of 0.8.0.16, every one of them a case
|
||||
**Bundled version: 0.8.1.0.** A table pass over 0.8.0.17, and a second-digit bump taken
|
||||
deliberately — 0.8.1 had been reserved for the seatless display table, which is getting more
|
||||
thought. **The one that changes state:** the Freight Agent's "clear the red Inbound box" returned
|
||||
the car to the Classification Yard still marked loaded, carrying a load already delivered and
|
||||
already paid for. `pooled()` strips a load's `origin` stamp and keeps `loaded` on purpose (a train
|
||||
can retire with freight aboard), so the call site's comment — "a car back in a yard is back in the
|
||||
common supply, carrying nothing" — described something it never did. Coaches suffered worst:
|
||||
detraining needs `coach && !loaded` out of the Division Yard, so a cleared coach came back as stock
|
||||
that could never unload a passenger. 18 loaded coaches against 6 empty over five measured games.
|
||||
|
||||
**Games in progress resume, and it was measured before release**, not argued from the diff: all
|
||||
twelve saves on the test box were replayed through `tryResumeSession` — the server's own boot check
|
||||
— against the new build. Six resume, six refuse, and the six refusals are the same six, at the same
|
||||
moves, with the same codes, that 0.8.0.17 already logged. **That replay is the check to repeat on
|
||||
any release that touches state**, because it answers before the install rather than after.
|
||||
|
||||
Also: leaving a running game no longer strands you in a lobby with no way to join or create
|
||||
(`runLobby` never restored the section `enterSeating` hides); the lobby's action buttons carry the
|
||||
board's amber; an Enhancement in the history names the Mainline card it went on, slot and type, by
|
||||
its printed name; the Manage Game dropdown says how long ago each game was last touched; and §6.2's
|
||||
Department refill — always implemented, never documented — is written into the game's rules
|
||||
reference.
|
||||
|
||||
**Previously, 0.8.0.17.** Four fixes from a live playtest of 0.8.0.16, every one of them a case
|
||||
of the game knowing something and the screen not saying it. **ABS Signals could be played on only
|
||||
one Mainline card** although its own tooltip said "any": the engine allowed all of them, but the
|
||||
action list drops duplicate labels and every placement described itself identically, so all but the
|
||||
|
||||
Reference in New Issue
Block a user