Commit graph

73 commits

Author SHA1 Message Date
Frederic Jung
1579f3743b
Merge pull request #311 from thesysadmindev/move-customer-path 2026-11-09 13:15:55 +00:00
Frederic Jung
249ad4d478
Merge pull request #320 from antt0n/perf/routing-core 2026-11-09 10:21:20 +00:00
Christopher Hawker
93d7c87dc7 Move a customer to another path between the same two POPs
Alien waves and leased fiber pairs sit on one fiber (b.link), so they move to
another working fiber between the same cities, under the rules for selling
capacity on it. Circuits have no link of their own, so moving one pins it to a
list of cities (b.pin), which routing follows while every link on it is up. When
one is down the circuit follows routing, and it comes back by itself.

The window lists the most diverse options first, compared by trench. Moves are
free and instant, and contracts keep their price.
2026-11-09 16:23:59 +11:00
Frederic Jung
4292cd4c74
Merge pull request #319 from antt0n/perf/intern-and-promises 2026-11-08 11:00:54 +00:00
Antoine
44116103be
The intern and customers' promises counted once per network: ports, caches and PNIs in one pass, shared searches
The game plays exactly as before.

* The NOC intern plans every POP in one pass: portsAt() counts router ports for all cities at once (routers, ports
  and cache orders; the ports job keeps ports in use and reserved by city between two jobs), cachePlan() sorts the
  boxes by POP once and cityShSum() sums a city's shares once per table, pniFlows() reads every POP's flows from
  routing's columns (R.flc) in one pass and pniWant() gives the weekly PNI round its sort key from typed columns.
  pniBatchRun() lets a frame through every few networks; pniQuote() looks existing PNIs and ports up from the batch.
* A dark fiber's capacity asks optG() once per run of identical optics (dfLit()); linkDown() and ownGlass() ask
  dfOn(), whether it carries anything for you, without the sum unless an optic is out or managed waves are sold on
  it. terrIslands() tests !l.sub first.
* Latency: latGraph() lists each city's links once in S.links order (typed arrays), latRun() is the same search with
  the same sums, latPairs() searches once per city for a boutique's finance routes and the latency offers.
* Diverse paths: divGraph()/divOn() number the cities and risks once; promiseCheck() shares a diversity graph, the
  land islands and one latency search per origin across every customer when the intern weighs taking a fiber out,
  and spots0() shares the graph for sovereign offers.

Tests: day-steps (cache planner, ports at every POP and flows from columns equal the old ways; the PNI round through
pniBatchRun), fiber-lit-and-latency (with failed optics and managed waves), shared-promise-checks, intern-personas.
2026-11-07 23:55:37 +01:00
Antoine
1edf0abb54
Routing core made once, list orders kept, and Auto-TE's detour searches cut: same routing, much less work
The game plays exactly as before.

* The core of compute() (rankPass, directPass, sendGroup, send, push, carry, altPath) is ROUTE in 29-route-core.js,
  built once by routeCore(): made inside compute() they were new functions every call and V8 threw away their code.
  Exits are read by number from typed columns (USED, CAPX, CIX, HPX, VISX), each list is an object of one shape.
* The route cache keeps, for each list and source city, the order routing tries it (xOrder()); a list whose exits,
  cities and preferences are unchanged on the same graph keeps its orders when other exits change (rcExitKeep()).
* Auto-TE: pairs with no detour are marked (TEF), exits that cannot help are skipped (dud()), one failed search
  answers the next ones, dijkstraTo() searches in arrays kept between calls, teSearch() runs on the graph as flat
  arrays and picks up where the last search from the same POP stopped.
* The PNI/IX tie-break (#298) reaches the core.

Tests: route-cache, routing-reference, speed (the fast core against the plain path, Auto-TE on crowded meshes with
POPs outside the core, picked-up searches counted).
2026-11-07 23:51:35 +01:00
Antoine
b9d6319302
The page stays free while big networks play: ambient map at 30 fps, saves in the worker, less to carry back
The game plays exactly as before; only what holds the page changes.

* mapStill() tells three cases apart: still (drawn once a second), ambient (only the animation clock moved: drawn at
  most every 29 ms, so 30 fps on 60 and 120 Hz screens) and moving (camera, pointer, selection, network: every frame).
  The animations follow the clock, so they keep their speed.
* save() skips the storage write when the text is the one the slot already holds (SLOT.txt) and no write is on its
  way; leaving the page still writes. A big autosave that moved on is packed, turned into text, stamped and
  compressed in the route worker (rwSave(), rwGzip()); the page only writes what comes back. A lost worker hands its
  pending saves back to the page (rwSaveBack(), rwGzBack()).
* Less crosses back from the worker: R.efg stays there while R is its own last answer (RW.wR, efg 0), R.egDesc comes
  as 0 or as the entries that changed (rwDescOut()/rwDescIn()), what crossed each link travels as typed arrays decoded
  on demand (rwLOut()/rwLIn()), and the full ports (hotMap) come along with R. Every message is read in order
  (rwMsg()), so both sides always hold the same last answer.
* Idle frames build the diverse-paths map and the latency floors a millisecond at a time (diversityWarm(),
  latWarm()), before the first offer asks.
* Diagnostic info shows how fast days really come: speed asked, speed held, how long a day takes (15-day-pace.js).
* perf-bench: "The NOC intern's shift: opsDay()" and, in Chromium, "One game day in play: page time only", a day
  played with the real route worker counting only the time it holds the page.

Tests: frame-pacing, day-pace, save-size, persist, route-worker (efg as typed arrays, hotMap from the worker),
day-steps (a second page plays the worker: answers without R.egDesc, with only its changes, posts without R.efg,
renewal haggles answered through the worker), intern-dark-fiber-worker, webgl-zoom-sprites, speed.
2026-11-07 22:22:51 +01:00
Frederic Jung
03523cb16b
Auto-release follows CI instead of testing the release again
A version bump on main ran the whole suite twice: once in CI for the
push, once more in auto-release. Auto-release now starts when CI
finishes on main and releases only if that run was green, tagging the
exact commit CI tested. Started by hand, it still tests first, since
there is no CI result to go by.
2026-11-07 10:00:15 +01:00
Frederic Jung
0342d90657
Faster tests, split everywhere: one sharded CI for pushes and releases
- The harness builds the DOM without the page's 2 MB of script text and
  runs the scripts straight in the window: a page load in tests takes
  about half the time (jsdom's parser was slow on the script text).
- That also took the game's source out of the DOM, which showed three
  tests that matched a phrase in the page source rather than the screen;
  they now check the panel, the routing window and the dictionary.
- CI plans its split once and runs the tests on 6 runners; every run on
  main records each file's time and caches it, so the parts stay even
  without anyone refreshing tests/durations.json.
- auto-release.yml and release.yml call the same sharded tests instead
  of running all 170 files on one runner each (11 and 15 min); a release
  is tested once, before it is tagged.
- Audit: five tests that duplicated a gameplay rule or checked nothing
  are gone; the two slowest checks sample a third of the cities, and a
  year of play to reach one due date became a moved due date.
2026-11-07 07:24:09 +01:00
Frederic Jung
52e740cdae
A slow phone gives a day longer slices between frames
A device that cannot hold 50 frames a second waited a whole slow frame at almost every rest of a big day, so the
game played up to a third slower at 8x. There a day gets half of its own frame, up to 32 ms; screens that keep up
keep 3 to 8 ms.
2026-11-06 22:48:52 +01:00
Frederic Jung
410eb81544
Merge pull request #313: every routing a day needs runs in the route worker, and a long day lets frames through
Conflicts with main resolved: the intern's chosen router model, optical upgrades and personas are generators like the
other orders, with a rest after each job. The optical upgrade's fiber test is a step the day answers, in the route
worker in play (opsDark(), {dark}), like the sales desk's trial; the renewal desk's reroute is a step too (slRenRun()).
2026-11-06 22:47:54 +01:00
Frederic Jung
74e8611f8e
The sales desk answers renewal haggles, each salesperson their way
With the new switch at the sales desk, whoever runs sales answers a
renewal haggle three days after it opens, unless the player did first.
The engineer gives business customers their discount and keeps nerdy
ones at the market price (nobody leaves); the shark holds the old price,
often with a fresh promise, bills a month of the saving, and loses some
(nerds most); the CEO gives 10% more off for a double term, says yes, or
picks a fight. Every answer is a hash of the contract. The switch is
stored only when on, so existing saves are unchanged.
2026-11-06 20:56:50 +01:00
Antoine
3c45d7dca1
A day that reroutes no longer stops the page at x8: every routing it needs goes to the route worker
The day split of #312 on its own, from 3.0.0, without its routing core, save or map changes. On a big network at
x8 the game stopped whenever a day rerouted: after a failure, an event, the NOC intern's shift or the sales desk's
signing, dayPost() routed on the page.

* The day is a generator (dayRun(), dayPreRun(), dayPost()): each yield is a compute() it needs. dayTick() answers
  them on the spot (tests, fallback); in play rwGo() posts each one to the route worker. A failure (failureRun()), a
  DDoS and Auto-RTBH (ddosNewsRun()), the chaos monkey's snip (mkSnipRun()), an event's cut (eventRun()), the
  intern's shift (opsRun()) and the sales desk (slAutoRun(), its trials answered by the worker with offerTrial())
  are steps of the day. fireFailure(), opsDay() and slAutoDay() still play them in one go.
* yield DAY_REST lets a long day draw a frame between two steps (half a frame, 3 to 8 ms, RW.budget). One day in
  flight at a time, resting or routing: the clock waits for the day, and a half day is never saved.
* UI.tick is set around every step (dayStep()), so quiet notifications hold in play, not only in dayTick().
* The daily mood check works out the land islands once per pass over the customers (TERR), instead of walking
  every land link for each customer afraid of sharks. This fix is thesysadmindev's, from #310; here the islands are
  worked out again if something routed meanwhile (RW.gen), since the mood loop can now rest between customers.

Same game either way: tests/day-steps.test.js plays a second page as the worker and compares each day with
dayTick(), the intern on every order, failures, events, DDoS and the monkey forced.

Thanks to thesysadmindev for the profiling in #310 that pointed at the reroutes, and for the sharks fix.
2026-11-06 14:24:18 +01:00
Frederic Jung
9a5522bbb1
Deals belong to Sales: the intern sticks to the network, the sales desk signs deals for you, orbit ones too
The NOC intern loses the sales order. Saves that had it load cleanly: sanitize drops the order, its log lines and its
weekly tally, and an intern who was running it hands its switch and minimum to the sales desk (S.slAuto).

With someone at the sales desk, the player can let them sign offers once a game day and decline the ones under a
minimum (39-sales-auto.js), each in their own style: the engineer and the CEO sign after the #185 routing trial (one a
day, a misfit waits a week), the shark signs every offer whose POPs are up, promises and commission included, and the
CEO's declines are a handshake away. Orbit offers need room on their landing dish instead of a trial. Nobody at the
desk (the default; the CEO is a choice, not a given): every deal is signed by hand. The sales desk runs before the
intern's shift, so a deal signed today is in the loads the intern works from. The sheet shows the switch, the minimum
and this week's and last week's tally; the NOC gets one notice a week.

Orbit offers go through the desk too (99-pie-sales.js): they carry o.by, so commission, renewal haggles and each
one's mark at renewal apply. The engineer only offers where the landing dish has 1.5x room on a well-served POP, and
those customers renew like the nerds (0.5x leave). The shark rings airlines, cruise lines and the Mars mission first,
at the biggest size that fits and 25% more, and half of them get a full-sky promise (REQS.skyfull: a crust at 90%)
before there is one; they grumble down to 40 happiness and never cancel over the promise alone. The CEO plays every
other golf game with the Founder: the biggest orbit deal that fits, Mars first, at 30% off; jitters cut his orbit
prices by 0.4x the jitters and from 0.6 the Founder skips golf. Without a salesperson orbit offers are drawn exactly
as before (the O7 balance save has nobody at the desk: payback unchanged, 7.0 years on seed 7, 7.5 on seed 11).

Also: the scenario reference plays 'sales' as the CEO signing for you; the route fuzz leaves out the state and
buttons of a standing order only main still has; the intern's on/off dials (paid peering, keep a spare) show their
translated value in the shipped page instead of the raw label.
2026-11-06 09:58:01 +01:00
Frederic Jung
2243280aff
Pie in the Sky: the Founder's antics make markets jumpy, then it wears off
- Every 60 to 120 days while a rocket company exists (the first 90 days after the plan starts), the
  Founder does something daft: a post, a gasp toast and a temporary effect (99-pie-jitters.js).
  A once-a-game antic takes orbit billing down for 3 to 5 days (no orbit income, then a restore post);
  a 3 a.m. spree (orbit offers x0.6 and a 4-point orbit mood dip for 14 days); a fight with a regulator
  (a landing-rights spat via pieSpatStart in a country with a dish or rights, within the spat limit);
  a leaked launch date (insurance 8% -> 14% for 30 days); Mars next year (offers x1.5 for 14 days, a
  promise like the others whose slip cancels the unhappiest 2% of orbit customers).
- Market jitters (S.pie.founder.jit, 0..1): each antic bumps it, it fades by 1/30 a day, and while
  high it lowers orbit offer odds (up to -30%) and holds orbit happiness under 100-25*jit (never low
  enough to cancel at renewal). The Company tab shows one line and a small meter next to the feed.
- Rule O5: harm only with pieChaos(); with chaos off he only posts. Dice only with a company.
- Saves: new founder fields in pieBlank, repaired in pieSanitize (pieAnticFix); old saves load calm.
- Insurance button shows the current premium (pie.ins_on_p replaces pie.ins_on). Agent docs updated.
- pie-balance: compute() also reruns when orbit billing goes down or comes back.
- tests/pie-founder.test.js: each antic, decay, once-only billing outage, chaos off, saves, 360 px.
2026-11-06 09:58:00 +01:00
Frederic Jung
9d74873edb
NOC tab and Pie Control tidied for phones: Pie Control in the NOC top row, traceroute in Routing & DDoS, government money in the bank
- NOC tab: Pie Control is a button in the top row next to "Open NOC dashboard" (only once the rocket
  company is unlocked) and replaces the Pie in the Sky card; a spat waiting for an answer keeps its card.
  The Connections button is gone (it lives in the NOC dashboard), and the traceroute moved into the
  Routing & DDoS modal as an entry that opens the unchanged traceroute modal.
- Government money (Rural Digital Pie Fund, Pitcairn grant, Ex-Im Pie Bank) is a section of the bank,
  shown with a rocket company or anything of the fund's to show. pieGovOpen, the action the toasts use,
  opens the bank there; pieTab({t:'gov'}) does too.
- Pie Control has four tabs (Traffic, Company, Launch, Dishes) in one segmented control that fits 360 px,
  styled like the HUD's speed buttons. The Founder's posts and the orbit books moved into the Company tab,
  newest three posts shown and the rest folded. A stale UI.pieTab of 'gov', 'book' or 'founder' falls back.
- Agent docs for pieGovOpen, pieTab, routingOpen and connOpen; the route fuzzer drops connOpen from the
  NOC tab on both pages to stay in lockstep with main.
- Tests: tests/noc-panel-layout.test.js, and the pie tests follow the moves.
2026-11-06 09:58:00 +01:00
Frederic Jung
8b92adcaa8
Pie in the Sky: government money for rocket companies
A Gov tab in Pie Control with three state programmes:

- Rural Digital Pie Fund: reverse auctions for towns only orbit reaches
  (a cut-off POP on backhaul, or a small town of your terminals). Ask
  90%, 70% or 50% of the reserve against a rival; the lower ask is paid
  monthly for five years while the town stays 70% served (monthly
  average) from a 90-day deadline. One bad month warns, two revoke the
  award and claw back half of the last year's support.
- Ex-Im Pie Bank: a BANKS loan of half the value of the pies flying and
  booked (at most 20% of a full Shortcrust, counted by O7), 1.5-2.5%
  over 60 months, never offered by the usual banks. Space covenants:
  launch n pies in a year, keep a pie dish in the lending country. A
  breach brings the bank's penalty, +3 points, a new launch deadline
  and, with chaos on, a "data stays home" or landing-rights spat there.
- Pitcairn: served from orbit (backhaul or home terminals) at 90% of its
  demand, latency forgiven, the UK pays a $120k/mo satellite grant.

Subsidies count in the orbit's books. tests/pie-gov.test.js covers the
features; pie-balance.test.js gains a run with government money, which
must pay back sooner by at most two years.
2026-11-06 09:57:51 +01:00
Frederic Jung
1de0723c35
Pie in the Sky: a full constellation pays back in about seven years
Rule O7's payback half is now played through the game loop
(tests/pie-balance.test.js): the big-network save runs a full lasered
Shortcrust by a sensible policy until its books balance, in 5 to 10 years
(6.3 on seed 7; 7.4 and 6.9 on seeds 11 and 3). The static formula is gone.

Retune: pies $0.5M (+$0.25M lasers), from $1M (+$0.5M); landing rights
$12,000 per eyeball up to $100M a year, from $28,000 up to $200M; orbit
rates x1.25 (pizza 25k, flywifi 36k, boat 27.5k, marsorb 44k); PIE_SELL
0.25 -> 0.4; PIE_OFFER_RATE 1/6 -> 1/3.
2026-11-06 09:57:51 +01:00
Frederic Jung
d7d5d21372
Pie in the Sky: dips in orbit hurt instead of emptying the sky
A crust carries a share of its traffic from 40% coverage, growing to all
of it at 60% (pieCarry/pieCap), instead of nothing until 60%. Pies wear out
over the last 30% of their life (pieAlive), deterministic, so a launch month
no longer deorbits on one day. Storms take 2-10% and food fights 5-10% of a
crust (PIE_HIT), down from up to 40% and 20%. Orbit customers lose at most
PIE_PATIENCE happiness a day without service, about a month to cancel.
2026-11-06 09:57:51 +01:00
Frederic Jung
940ce0df10
Pie in the Sky: backhaul for POPs cut off from the backbone
A pie dish at a POP the backbone does not reach now lifts all of that
POP's own traffic to a landing dish on the backbone (bent pipe within
1,000 km, lasers past that): one virtual edge, capacity min(dish, landing
dish, crust) after terminals, latency the orbit path. Its IGP weight is
Infinity; only the cut-off POP's paths are rerun with it (pieRelink), so
POPs on the backbone never route through orbit and saves without a
cut-off dish route exactly as before. Spats, safe mode, a down crust,
landing rights and backhoes cut it.

A cut-off dish now reports 'backhaul' or 'isl' (it said 'ok' when it
had local terminals). Rule O3 now reads: a cut-off pie dish carries only
its own POP's traffic and local terminals, never transit between other
cities.
2026-11-06 09:57:40 +01:00
Frederic Jung
d2c7294ed3
Plan: Pie in the Sky constellation spec 2026-11-06 09:57:23 +01:00
Frederic Jung
f303f29f22
PNIs and IX ports are both peering: traffic leaves at the nearest one (#298)
Routing tried every PNI (rank 1) before any IX port (rank 2), like a
local-pref above peering, so a PNI in London pulled New York's traffic
across the Atlantic even with the same network at the New York IX. PNIs
now route at rank 2 with the IX ports, nearest first. At the same distance
the PNI goes first: an IX port counts as PNI_TIE (0.001) further away,
far below any link weight (at least 1). Distances compare as
(Da-Db)+(dta-dtb) so exits at one distance stay exactly equal, and the
POP-local shortcut skips lists that mix PNIs and IX ports. Backup only and
RouteScript's prefer pni keep their say. The traceroute, the reference
allocation in the tests and the RouteScript help follow.
2026-11-06 06:31:21 +00:00
Frederic Jung
98735eb769
Container image: releases only, a token by default, lean layers; every action pinned
The image is published to GHCR as the release's version when a release
dispatches docker.yml, and never as latest or per commit. Pushes and pull
requests that touch it build it and try headless (with and without a
token) and browser mode, without publishing.

The image: node:24-slim pinned by digest (the major CI tests on), npm ci
from the lockfile in its own layer, only src, site, scripts and the
harness copied in (an allow-list .dockerignore), root-owned and run as
node, /saves ready for --save-dir. The entrypoint runs the servers with
node directly and, with no SOT_MCP_TOKEN, makes a token for the run and
prints it (MCP_NO_TOKEN=1 to opt out); the docs publish ports to
127.0.0.1 and explain --allow-host. Labels, SBOM and provenance on push.

Every GitHub Action in every workflow is pinned to a commit SHA with its
version, and Dependabot keeps the actions, the image's base and the npm
tools current. tests/docker.test.js guards all of it. The ADR and the
github-pages skill say the image is a tool: GHCR only, never Pages.
2026-11-05 20:14:04 +01:00
Wolfgang Nagele
9c8cda44c1
Build container images for easy MCP use 2026-11-05 20:14:04 +01:00
thesysadmindev
e62e4c7c98
Merge branch 'main' into unbuilt-pops 2026-11-05 21:01:24 +11:00
Frederic Jung
13768c61cb
Merge #281: draw the map with WebGL2 by default, 2D as fallback 2026-11-05 10:13:28 +01:00
Frederic Jung
27c4977412
A sales desk: the engineer turned salesdroid, the shark, or the CEO
From customer level 2, one person runs sales at a time, next to the NOC intern:
- The engineer turned salesdroid ($9k/mo) books you into every network operators' meeting 30 days
  ahead at the usual ticket price, brings 40% more offers (half as many again while a meeting's glow
  lasts) but never the huge deals, and their customers leave less often at renewal (half as often
  for the nerdy ones: HFT, ISPs, universities, L2). Haggles badly.
- The shark ($15k/mo) brings bigger sizes at 25% more, takes a month of each deal as commission,
  oversells half of them with a requirement the network does not meet yet, never brings technical
  customers, makes selective peering desks want 25% more traffic, wins most haggles, and their
  customers leave 50% more often.
- The CEO (free) signs the biggest internet customer the network can take on the golf course every
  50-80 days at 30% off, often with a promise nobody checked, signs about a third of the offers you
  decline anyway, and has opinions. He cannot be fired: a board meeting brings a VP of Sales a month
  later.
Customers remember who signed them (by), which the renewals read. Every roll is a hash of the day
or the deal, so games without a salesperson keep their random stream.

The differential fuzz also leaves out buttons for actions main does not have yet.
2026-11-05 09:32:24 +01:00
Christopher Hawker
5259e93034 Merge remote-tracking branch 'upstream/main' into unbuilt-pops
# Conflicts:
#	CHANGELOG.md
2026-11-05 18:27:21 +11:00
Christopher Hawker
0cdbef97fe A dashboard tab listing the cities with no POP yet
The dashboard shows what you have built, not what you haven't. The new tab lists every city you could still open, with its eyeballs, exchange and price, and sorts, filters by region and pages like the other long lists. Open selects the city on the map, so a build still goes through the city panel and its warnings. The tab also works in the NOC window.
2026-11-05 18:26:44 +11:00
Frederic Jung
8841fdacb9
Customers come and go: contracts, renewals at the market price, growth, haggles and leaving (#178)
Every customer now signs for a term (enterprise and hosting 1-3 years, wholesale and circuits 2-5,
fiber pairs 10-20). When it ends most renew at today's bandwidth price, happy ones often grow to
the next size, a few with competitors around ask for a discount 30 days ahead (match, meet halfway
or let go in Deals; no answer matches), and some leave: unhappy ones because of the service, others
acquired, bust, insourced, moved (they stay if you have a POP there), multihomed, project ended,
a cheaper bid, or now and then one of the odder reasons (never more than one every 60 days).

The bandwidth market is flat for six months, then about 10% cheaper a year, slowed by dearer
hardware, moving in a slow three-year cycle and pushed up during a shortage. New offers and
renewals are priced in it; one renewal moves a price at most 25% down or 10% up.

Nothing happens in a first term. Rolls are hashes of the customer and its term, never Math.random,
so seeded games keep their stream. Old saves get renewals spread over the next year. Happy
customers of a year or more send a referral about every six weeks. The dashboard's Customers tab
shows customers won and lost each month and why, renewals due and the market; customer rows show
the contract end.

The differential fuzz now leaves out state fields only this checkout has, so a new feature's state
does not fail the comparison with main while everything both games share must still match.
2026-11-05 08:15:53 +01:00
Antoine
cf9ab5e774
Draw the map with WebGL2 by default, with the 2D canvas as fallback
- Sea, land, waves, clouds, cables, tubes and packets are drawn by the
  graphics card; the land is geometry built once, so panning and zooming
  repaint nothing.
- Cities, labels and markers are painted once into a sprite atlas and drawn
  in one call (with drawImage in 2D, which speeds up the 2D map too).
- Settings > Map drawing: Auto or Always 2D, per device. Browsers without
  WebGL2, a lost context and puzzles keep the 2D map.
- npm run bench:gpu compares 2D and WebGL in a real browser on a real GPU;
  ?fps shows a frame meter.
2026-11-04 22:37:12 +01:00
Christopher Hawker
f9d24699c1 Merge remote-tracking branch 'upstream/main' into noc-window
# Conflicts:
#	CHANGELOG.md
#	src/game/99-loop.js
#	tests/harness.js
2026-11-04 20:01:57 +11:00
Christopher Hawker
80f671953a A NOC screen in its own window, for a second display
A button next to the speed controls opens this page at ?noc=<slot> in a
noopener pop-up, so it runs on its own thread. The window never claims the
slot, saves or runs a day. The game sends it a copy of S and R over a
BroadcastChannel, at most once a second and only while a window is connected,
and runs its clicks through an allowlist of the same ACT.* a click uses.

The window opens on a dark wallboard: a live world map of links and POPs,
traffic and money graphs, alarms, and the hottest ports with the dashboard's
own upgrade buttons (connUp). Clicking a POP filters the ports. Details shows
the full dashboard. Phones do not show the button.
2026-11-04 20:01:37 +11:00
Wolfgang Nagele
7438fd1065 Agent mode: an AI agent plays in the player's own tab, with the agent's clock or a running one
The MCP server gains a second session, --browser: the game in the player's tab, reached through the
agent relay npm run dev hosts at /agent (loopback unless --host; Host and Origin checked, no token).
The player turns it on in Settings → Agent API, a section only the development page has, and picks the
clock: a pace of 30 s to 5 min a day that nobody can pause, or the agent's own. Headless gets the
running clock too (--clock realtime --pace MS) and both get a wait tool.

All of it lives in src/agent/ (development page only). The game offers HOOKS, UI.clockLocked,
setSpeed() and runClock(dt), the frame's day block moved out unchanged, and nothing more: the page
players get grows by under 1 KB gzipped and runs exactly as before.
2026-11-03 16:18:43 -05:00
Wolfgang Nagele
2526dd458d MCP server on the official SDK, over stdio or HTTP
scripts/mcp.js registers the same tools on @modelcontextprotocol/server (2.4.0, with
@modelcontextprotocol/node 2.1.2 for node:http; dev dependencies, pinned) instead of its own JSON-RPC
loop. stdio works as before; --http [HOST:]PORT serves Streamable HTTP at /mcp, on loopback unless
told otherwise.

Over HTTP every request's Host must be an address of this machine (or --allow-host), an Origin other
than localhost is refused, and a bearer token (SOT_MCP_TOKEN or --token-file) is optional; bound
beyond loopback without one, the server warns. load_save and export_save take JSON only there, or files
under --save-dir and nowhere else. Calls from every client run one at a time, in order.
2026-11-03 16:18:38 -05:00
Wolfgang Nagele
5b76685e5c MCP: the agent facade moves into src/agent/, assembled into the development page only
The command tables and the page-side reads, screens and dispatch (scripts/lib/mcp-commands.js and
mcp-page.js) become src/agent/, the facade window.TUBES.call(tool, args) in the page's own scope.
A new <!-- @include-dev agent --> line in src/index.html pastes it in after the game, but only in the
development page (npm run dev, the MCP server, i18n-check, tests that ask with load({agent:true})).
The build drops the line whole: _site/ is byte for byte what main builds.

Tests load the page players get by default, the build test fails if anything outside src/agent/ names
the facade's api/API_ names, and the pages test checks the built page carries none of it and still
plays. ADR 0004 records the rule.
2026-11-03 16:18:38 -05:00
Frederic Jung
83740911e8
MCP: new_game takes the Best Effort preset (soft); docs and the rules reference say so 2026-11-03 10:22:30 +00:00
Wolfgang Nagele
a72cb7d760 MCP server: typed, checked actions with outcomes and dry-run prices; schema, catalog, state, results and ui tools 2026-11-02 21:10:35 -05:00
Frederic Jung
640b65ce9a
Differential random games: main's page and this checkout's play the same seeded game move for move
scripts/route-fuzz.js plays many random games (every mode: presets, sandbox, scenarios,
the weekly, puzzles, architect runs, or a save) on two pages in one process and compares
all of R, R.fl, R.ef, the PNI lists and S after every move and every day, and routing with
the kept tables against routing without them. tests/route-fuzz.test.js plays four.
2026-11-01 21:44:30 +01:00
Frederic Jung
8b0c624b21
Merge remote-tracking branch 'origin/main' into perf/6-routing-design 2026-11-01 19:42:12 +01:00
Frederic Jung
37def9dd6b
The game as a local MCP server: play, inspect and debug it from an AI client
scripts/mcp.js speaks MCP over stdio and loads the game from src/ in jsdom,
seeded, like the tests. Tools: new_game, load_save, export_save, status,
advance (with per-day timings), network, city, find, traceroute,
peering_check, hot_spots, actions, act, screen, panel, click, open_link.
Actions go through ACT as buttons do; files, slots, settings and the clock
stay out of reach, and nothing evaluates code in the page. Nothing is
hosted and the page is unchanged. See docs/MCP.md.
2026-11-01 18:04:22 +00:00
Frederic Jung
2122db8cc4
Architecture notes on the kept routing tables and compute()'s small functions; CHANGELOG 2026-11-01 19:04:18 +01:00
Frederic Jung
18db953818
Merge remote-tracking branch 'origin/main' into perf/3-worker
# Conflicts:
#	src/game/80-simulation/30-compute.js
2026-11-01 16:14:07 +01:00
Frederic Jung
96e2c2f537
Cable ships: club repair ships sail out to subsea cuts, private charters jump the queue, lay ships build your cables
A subsea cut sends the nearest free ship of its region's maintenance club from its port, along the cable routes at about 500 km a day; the cut lasts while it sails, waits for permits in slow countries and splices (3-5 days shallow, 6-12 deep). Ships stay busy until home (S.fleet). With the club's ships all out the cut waits; a private charter of the quickest free ship costs $1M + $200/km sailed. Own subsea builds show survey/factory, then a lay ship at 150 km/day; a second lay ship can be booked before laying ($3M + $150k/day saved), and crews no longer speed up cables at sea. Ships are drawn on the map and can be tapped.
2026-11-01 11:24:53 +00:00
Christopher Hawker
eb78d7104e Prefer keeps the say at its own POP only, and the direct ranks stop slowing the game day
Review of #205:

- Prefer no longer switches the carrier-direct rule off for the whole network.
  A POP with a Prefer-marked transit port keeps what Prefer says; every other
  POP keeps the rule. compute(), traceroute and the reference loop in the tests
  all use the same per-POP rule, and a new test covers it (New York and Chicago,
  Prefer on one POP's Tell-Ya port).
- dayTick() cost. The harness is neutral (the same page on both sides differs by
  1%), but the cost was real: the four extra rank passes inside compute()'s main
  loop made V8 deoptimise compute() 954 times in 30 game days against 46 on main.
  The direct ranks now run in a small function of their own, in the plain way the
  reference loop does (their lists are one carrier's ports), so the main loop is
  the one main has, and the setup moved out of compute(). 61 deopts instead of 954.
- Italian: "Porta qui" read as "port here" next to "Porta più vicina"; it is
  "Instrada qui" now, and "Ogni POP instrada il traffico..." matches.
- Help text: says that real BGP carries the traffic from every POP (the shorter
  AS path beats IGP distance), that Carry here does that here, and that Nearest
  port is the default. All six languages.
2026-11-01 20:07:23 +11:00
Christopher Hawker
9dc8aac45a Merge remote-tracking branch 'upstream/main' into own-carrier-egress
# Conflicts:
#	CHANGELOG.md
#	src/game/80-simulation/30-compute.js
#	tests/game.test.js
2026-10-31 23:57:19 +11:00
Frederic Jung
72fb233af7
DDoS rework (#242): world-sized floods on customers, spread over every port, RTBH per target, Flowspec rule budgets, magnet scrubbing, new NOC card
Work in progress: the full test suite is still being brought in line.
2026-10-31 12:20:14 +01:00
thesysadmindev
296de1d946
Merge branch 'main' into own-carrier-egress 2026-10-31 22:16:55 +11:00
Frederic Jung
92536606dd
Tests, CHANGELOG and docs for the peering desk 2026-10-29 21:41:28 +01:00
Frederic Jung
0ece48a79d
Ship weekly scenarios with the weekly puzzle run; docs and skills 2026-10-29 17:34:38 +01:00