Skip to content

Releases

v4.42.0

Choose a tag to compare

@github-actions github-actions released this 14 Nov 08:47
  • Fixed: heavy lag from the Auto buyer, worst right after an ascension. A rule whose condition reads
    buyer.budget() (the seeded "Buy all upgrades → buy first when 20 or more" rule) worked the whole queue out again
    inside itself, over and over, every time the buyer looked; inside its own queue buyer.budget() is now the bank
    less the reserves placed first.
  • Auto buyer: while a building's next 100 cost under the profile's Cheap ×100 seconds of CpS, the queue is just
    those (cheapest first, plus Buy all upgrades when a rule puts it first) and nothing else is measured: right after
    an ascension it buys hundreds of them in a row without computing a payback period for each.
  • Faster: the "to next achievement" count only tries the counts that can win an achievement (the same answers).

v4.41.1

Choose a tag to compare

@github-actions github-actions released this 14 Nov 08:15
  • Guide: the Auto buyer section covers the To next achievement profile setting.

v4.41.0

Choose a tag to compare

@github-actions github-actions released this 14 Nov 08:12
  • Auto buyer, To next achievement (profile setting, on by default): besides its bulk, each building is also
    measured bought up to the next achievement it wins (within 100), and the queue takes that count when its
    payback period is better. A building a few short of a tier now gets credit for the milk that tier brings,
    spread over the whole buy. Its queue row carries a "to next achievement" badge.

v4.40.2

Choose a tag to compare

@github-actions github-actions released this 12 Nov 01:27
  • American English everywhere: the last British spellings (colour, favourite, customise, behaviour, neighbour…) are gone from the add-on's text, code, docs and tests. Old names keep working: saves with the storeColours option, statistics saved or shared with a colour field, and alert.flash(colour=…) in macros all read as before.

Simulator 0.8.1

  • American English: the garden model's neighbours() is now neighbors().

sim-v0.9.0

Choose a tag to compare

@github-actions github-actions released this 12 Nov 11:47
  • Golden cookie income, computed: a save's income curve now takes seconds instead of minutes, and its points
    are within a few tenths of a percent instead of ±5–25%. Golden cookies are a renewal process (the spawn timer
    doesn't care what's running) and their effects a Markov chain over what's running (each buff's time left, the
    last effect), so models/goldenExpected.mjs walks that chain a million golden cookies a frequency and credits
    each one its expected payoff: every effect it could pick (with the game's own odds from popFunc, Dragonflight
    and the last effect included), what it pays, and what CpS, clicks and storm drops earn until the next one,
    averaged over the wait's exact distribution. Rare stacks (Click frenzy or Dragonflight on Frenzy, Dragon Harvest
    and building specials) are then counted at their odds every time their makings come up, instead of the few
    times the dice land on them. exactIncome solves the same chain outright on a time grid (no noise, but only
    up to +50% or so before its states run away): the walk agrees with it within 0.1%. Checked against the game
    played event to event: golden cookies and effects an hour, the time each mix of buffs runs, CpS and Lucky
    payouts.
  • Score for your game (Garden page) computes the curve by default; Played 2,000 h / 8,000 h still play it.
    A computed curve goes through its points (a smooth monotone interpolation) instead of a fit. The bank is held
    at the save's (Lucky and chains read it); golden-income --bank sets another. golden-income and
    garden-score compute too (--pops), or play with --play.
  • Following a frequency over time: the computed walk can follow a golden cookie frequency (and CpS factor)
    that changes over game time, the timer set at each golden cookie as the game sets it. garden-score --follow
    scores each gardener that way too, from its own frequency over time, as a check on the curve (which treats the
    frequency as steady at each moment): on a late-game save they agree within about 1% for the gardeners worth
    having. The computed model is now the simulator's default for golden cookie income; playing (--play) checks it.
  • Fixed, the played income model: a free sugar lump no longer lifts CpS for the rest of a run (with Sugar
    baking each one added 1% against a baseline taken at the start: ~4% too much over 2,000 h), and the 1% second
    golden cookie now sees the first one's buff (it used CpS from before it).

sim-v0.8.1

Choose a tag to compare

@github-actions github-actions released this 12 Nov 01:27
  • American English everywhere: the last British spellings (colour, favourite, customise, behaviour, neighbour…) are gone from the add-on's text, code, docs and tests. Old names keep working: saves with the storeColours option, statistics saved or shared with a colour field, and alert.flash(colour=…) in macros all read as before.

Simulator 0.8.1

  • American English: the garden model's neighbours() is now neighbors().

sim-v0.8.0

Choose a tag to compare

@github-actions github-actions released this 12 Nov 00:23
  • Score garden strategies for your game: the Garden page's new Score for your game card takes a save and
    ranks each gardener by the income it implies, in multiples of a baseline (unbuffed CpS + clicks a second), with
    golden cookies clicked as they appear. The save's income curve comes first: the game's own golden cookies played
    at steady frequencies for thousands of hours each (models/goldenIncome.mjs, jumping from one event to the next
    instead of playing every frame), smoothed by a fit; then each gardener's time at each frequency, its CpS cost
    included, goes through it (models/gardenScore.mjs). Also as experiments: golden-income, garden-score.
  • A save as a game state: the few variables golden cookie income depends on, read from a save and each one
    changeable: buildings, clicks a second, a click's share of CpS, dragon auras, Grandmapocalypse, season, Pantheon
    and the golden cookie upgrades (the card's form; golden-state, --state file.json, --set key=value).
  • Garden model in step with the add-on: clover rounds switch to clay once a quarter of them are mature (was 75%),
    as the add-on's soil rules have since PR #131.

v4.40.1

Choose a tag to compare

@github-actions github-actions released this 11 Nov 23:28
  • README: the Stock market tile shows the Trader card and names the trader profiles; screenshots retaken after the trimmed captions.

v4.40.0

Choose a tag to compare

@github-actions github-actions released this 11 Nov 23:21
  • Fewer labels: captions, hint lines and kicker labels that only repeated what a card, chart or control already
    shows are gone or shorter, across every page, widget and tooltip: the Auto buyer, Queue, Purchase history and
    profile cards; the Buildings and Upgrades cards; the Garden's slots and editors; the Pantheon, Dragon and
    Grimoire pages; the macro, statistic and code editors; the Managers page; the Widgets page and widget settings;
    the Stats cards and chart descriptions; and the chart hover readout. Explanations you still need moved into the
    hover tips, and table headers, stat labels and slot names stay.
  • Widget popups: no more "click to open…" line at the bottom; clicking still opens the page.

v4.39.0

Choose a tag to compare

@github-actions github-actions released this 11 Nov 23:18
  • Stock trader profiles: the Stock market autobuyer runs a trader profile, code you load and unload like the
    garden, Pantheon and Grimoire profiles — on the Stock market page's new Trader card (its thresholds as knobs,
    Save, Save as new, Revert, rename, delete), from code (stocks.profile(p), stocks.profileActive(p),
    stocks.activeProfile(), stocks.profiles()) and from a manager's stage (a Stock trader card).
  • Trends and prices, the new default trader: fast rises bought and held, fast falls sold, every other trend
    bought under its resting value and sold over it, by thresholds a simulator study found — about 60% more profit
    than Ride the trends (the strategy until now). A save from before keeps Ride the trends; autobuyer code you
    had edited becomes a profile of yours.

Simulator 0.7.0

  • experiment stock-market: the Stock market autobuyer's trader profiles in the game's own market code, run as
    the add-on runs them: profit per hour, what a market tick spends (gross and net, by percentile), the capital in
    stocks, and how much of the profit a bank keeping only so much for stocks still makes. --search finds Trends and
    prices' thresholds (price paths from the game, a coordinate search, checked on held-out seeds).