The generated cart release workflow curls cartkit.py into RUNNER_TEMP and
runs the selftest from the cart repo, where find_repo() finds nothing and
scaffold exits non-zero, so every cart release failed at the first step.
Report failures and skips even under --quiet; the workflow passes it, so
the only output was the FAIL count with no indication of which check.
PcMenu.lua's own CHANGE BOX save confirmation stayed in English on every
non-English build: its overwrite/saving/done prompts and its YES/NO choice
were bare literals, invisible to a translation mod's `strings` registry,
even though the overwrite/saving prompts are the exact same two cart
messages Gold's SAVE screen (SaveMenu.lua) already routes through
Strings().
SaveMenu.lua now exports OVERWRITE_PROMPT_SOURCE, SAVING_PROMPT_SOURCE and
twoLines() so PcMenu:savePrompt() can call through them instead of
duplicating the two shared prompts. The now-unused plain
OVERWRITE_PROMPT/SAVING_PROMPT tables (PcMenu's old direct-read contract)
are gone along with them, removing a second copy of the same literal text
that had to be kept in sync by hand. The CHANGE BOX confirm prompt itself
("#MON BOX, data\nwill be saved. OK?") gets its own new Strings.source()
key -- it is not shared with SaveMenu, and the port's own English text is
already a same-shape truncation of the cart's three-line
_ChangeBoxSaveText, confirmed against poke-corpus GoldSilver
en_msg.txt:4897. The done message and YES/NO now go through the exact
same Strings() keys SaveMenu.lua already uses ("%s saved\nthe game.",
"Could not save.", "YES", "NO"), so a translation covering SaveMenu's
screen needs no PcMenu-specific fork for any of those.
Added tests/engine/gen2_pcmenu_changebox_save_translation_test.lua,
mirroring gen2_save_menu_translation_test.lua's technique for PcMenu's own
CHANGE BOX flow, and updated the stale cross-file-contract regression case
in gen2_save_menu_translation_test.lua to match the new shared-export
shape instead of the old plain-table one.
An independent review flagged two gaps:
Pokegear.lua's clock card (drawClock and the card-list clock view) prints
the same AM/PM information MainMenu's clock box does, but was left as bare
literals -- so a mod translating AM/PM would now show it correctly on the
title screen and in English on the Pokegear for the exact same hour. Both
call sites now go through Strings("AM")/Strings("PM"), the same source key
MainMenu.lua already added, so no new corpus entry is needed on the
translation-mods side.
SaveMenu.lua's twoLines() only ever split on the first "\n": drawPanel's
box has room for exactly two Chrome.print calls, so a translation needing
a third line (nothing in the current fr/de/es/it overrides does, but
nothing stopped one from trying) had nowhere on screen to go, and the
extra "\n" landed as a literal newline byte drawn as glyph garbage on the
second line with no indication anything was wrong. It now warns once per
source text instead.
Added a gen2_save_menu_translation_test.lua case driving drawPanel() with a
three-line mod translation: confirms only the first two lines render, the
warning fires exactly once (not once per frame), and drawing does not
crash. Pokegear.lua has no engine-tier test suite to extend for the AM/PM
change; verified by direct inspection and a syntax check instead.
MainMenu.lua's row labels (CONTINUE/NEW GAME/OPTION/EXIT GAME), its clock
box's AM/PM half, and its CONTINUE save-summary panel (PLAYER <name>/
BADGES/POKéDEX/TIME, or NO SAVE FILE) were bare literals, invisible to a
translation mod's strings registry. SaveMenu.lua's confirm/overwrite/
saving/saved prompts, its YES/NO choice, and its own copy of the same
summary panel had the same gap. Both mirror the Gen 1 port's already-
translated equivalents (src/ui/TitleState.lua, src/ui/StartMenu.lua),
which route every one of these rows through Strings().
SaveMenu's two-line prompts (module-level OVERWRITE_PROMPT/SAVING_PROMPT,
plus the dynamically-built "%s saved\nthe game." and the confirm/failed
messages) are now single Strings()-resolved strings with an embedded "\n",
matching the Gen 1 port's own single-call convention for two-line messages,
split into the two-slot table drawPanel's fixed-position Chrome.print calls
expect only at draw time -- so a translation sees one whole sentence to
reorder, not two independently-translated fragments.
Added tests/engine/gen2_main_menu_translation_test.lua and
gen2_save_menu_translation_test.lua: drive both screens' drawPanel()/
drawSavePanel() with a mod-loaded Strings catalog and check the translated
text reaches Font.draw, plus a vanilla no-mod case proving the fallback is
unchanged. Confirmed both catch the regression: reverting either file to
its pre-fix content fails the corresponding suite (7 and 15 checks
respectively).
Both features landed independently and each was green on its own, so the gap
only appears in the combination.
- tools/cartkit.py BASES did not list crystal, so cartkit rejected a Crystal
cart as an unknown base game. The Lua validator never had this problem: it
asks GameVersion.VERSIONS, so it accepted Crystal the moment the row existed.
- Two self-tests used the literal "crystal" as their never-a-game sentinel and
invert now that Crystal is real: tests/engine/cart_manifest.lua and
cartkit's own t_identity. Both re-pointed at "nonesuch", the token the rest
of the suite already uses for this.
A custom cart pairs an identity (title, shell colour, label art) with a base
game, a list of mods pinned to exact builds with their option values frozen,
a load order, and a seal. It ships no code of its own: every mod it names is
a separately published mod, which is what keeps a cart auditable before it
runs and reproducible after an author's repo disappears.
Format and storage:
- src/carts/CartManifest.lua parses and validates cart.json, canonicalises it
for hashing and reads/writes the .g1rcart bundle. The bundle is a data-only
serialised table read through SaveSerializer, so an imported cart can never
execute code. Canonical strings are length-prefixed because option keys and
values are author-controlled and could otherwise forge a record boundary and
collide two different carts onto one hash.
- Pins name a public source: a GitHub release with its sha256, a GameBanana
file id with its md5, or "local" for a capture that only exists on this
install. A local pin is unpublishable by construction, which is what makes
"build it here, publish later" possible without inventing a hash.
- Label art rides alongside the manifest rather than inside its identity, so
re-arting a cart does not tell every player their run is out of date.
src/core/Base64.lua decodes it; strict, with no whitespace tolerance.
Saves:
- Cart playthroughs live in the cart's own slot namespace (saves/cart_<id>/),
so a cart's file never sits beside a vanilla one and uninstalling a cart
never orphans a save. Every save records the cart build it was made under.
The seal:
- A sealed cart loads its pinned list, in its order, with its options, and
nothing else. A pinned mod with no frozen options gets an empty bucket so
unfrozen keys fall to schema defaults, identical for everyone; otherwise two
players on one cart quietly run different games.
- A sealed cart refuses to load when a pin is missing or installed at another
version. Playing a subset of the cart is the exact dishonesty the seal
exists to prevent, so the refusal loads nothing at all.
- Breaking the seal is permanent, marked per save slot, and downgrades that
playthrough to open behaviour. It cannot be cleared through any public API.
Launcher:
- A game's page carries a Custom Carts control and a picker; choosing a cart
turns the page into that cart's page, with its own cartridge, title and save
slots. The rail of five games never grows and a cart id never reaches
imp.tab or imp.panelVersion.
- Loader.planCart runs before boot so a refusal is visible on the page instead
of being discovered as an error after launch.
- Save as cart captures the enabled mods for a game and names, before the
player confirms, every mod that could only be pinned to this install and
whether the result can be shared at all.
Authoring:
- tools/cartkit.py scaffolds, validates, pins and packs a cart, and installs a
release workflow. Its writer is byte-identical to the engine's serialiser.
Crystal boots from a user-supplied ROM, imports a full cache and is playable:
copyright, the Crystal intro movie, the animated title, gender select, Oak,
and out into Johto. 122 of the cart's 169 script specials are implemented.
Import and data
- tools/make_crystal_manifest.py derives the manifest by importing
make_gold_manifest as a library, with three additive keyword seams. Gold and
Silver still regenerate byte-identical, which is the standing requirement for
touching that generator.
- crystal_symbol_deltas.py and crystal_movie_symbols.py carry the symbol delta:
Crystal renames the credits mons, splits the trainer card, Pokegear and
pack-pal blocks by gender, and replaces the intro and title outright.
- Crystal-only manifest keys: engineFlagOrder (162 flags to Gold's 93, so the
badge block sits one higher) and unownCharmap (the main charmap parser stops
at the first newcharmap so the two cannot contaminate each other).
Extractor
- RomExtractorGen2 becomes three-edition. Crystal corrections: PAL_MAP_BANK
0x13, a flat PICS_FIX pic bank, audio bank 0x5e, the mapSongs id-100 hole,
seven NPC trades, a TradeTexts stride of 8, the five Crystal tileset anim
steps with per-row degrade, and the column-major trainer card portraits.
- New: animated front sprites (frames, bitmasks, play and idle scripts), the
Battle Tower roster, Kris assets, Mobile System GB art, and the Crystal
intro and title via src/import/CrystalMovie.lua.
Engine
- GameVersion gains engine(id) and fixes(id). Gold and Silver keep their
original bugs where the bug is not hardware dependent; Crystal gets the fixes
Crystal shipped: Lucky Number boxes 10-14, surfing onto an NPC, and the
Reflect and Light Screen defence overflow.
- Crystal story: Suicune and Eusine, Celebi behind the GS Ball flag, the Ruins
of Alph chambers, Buena, the Move Tutor, the Poke Seer, and the Battle Tower
including the wInBattleTowerBattle badge-boost guard.
- Kris and the gender flag, animated fronts in battle and the summary screen,
and mon caught data.
Verification
- Every extracted asset is pixel-compared against pret's own source PNGs.
- Gold caches are byte-identical before and after, file for file.
- New Crystal suites plus a T2 Gen 2 tier; the full suite passes.
- Replaced `teardownMountedSession` and `flushEditorPackageLoaded` with a unified `SessionLifecycle` approach for managing session transitions and resource cleanup.
- Implemented `SessionLifecycle.endEditorSession` and `SessionLifecycle.endGameSession` to streamline the teardown process for editor and game sessions.
- Introduced `Assets.releaseSession` to handle GPU resource release at session end, ensuring efficient memory management.
- Updated `Game` and `Game2` reset methods to include world and canvas resource releases.
- Enhanced `MapLoader` with a new `releaseAll` method for eager GPU cache cleanup.
- Added tests to verify the new session lifecycle functionality and resource management.
This commit improves the stability and performance of in-process transitions, particularly during editor and game session changes.
further improves and addresses #1662 specifically around android gc pressure.
Route B in CONTRIBUTING-mods.md asks an event/hook change for five things.
This adds the two that were missing and fixes what the third turned up.
link.battle_ended built its payload unconditionally. Route B is explicit
that a new event must not allocate when nothing wants it, and every other
emit in the engine already guards -- Runtime.wants now gates this one too,
so an unsubscribed build runs the branch exactly as it did.
world.talk was handing Runtime.call a closure built fresh on every A press,
purely to have a fallthrough to pass. It is a file-local now, so an unhooked
press allocates nothing it did not allocate before.
The RFC covers motivation, the API delta with call sites, migration (nothing
changes for existing mods), and verification. The backward-compatibility
statement is in it: every item is a new name or a new optional argument, and
example_mew_starter -- api 1, category = "GAMEPLAY", whole-species copy --
still loads, which run_modkit proves on every run.
No registry or schema field is added, so gen_registry_docs has nothing to
emit for this change. Running it does show pre-existing drift in
docs/modding/reference/registries.md (timeFishGroups, an objects refinement)
from earlier Schemas.lua edits that were never regenerated; that is not this
branch's to carry, so it is left alone.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
gate_meta_coverage asks every extension point for a unit test through the
public mod API, a no-mod parity test, and docs. The seams commit shipped the
call sites and owed the rest.
Both cases drive the real engine path, and both assert the unhooked build
first, so they fail if the seam is deleted and if it changes vanilla
behaviour. world_talk stands an object on the faced cell: with no mod the A
press reaches talkTo, with a mod that owns the object it does not, and an
object the mod ignores still falls through. link_battle_ended parks a
session at the end of a battle and checks the event carries the result, the
role, and both party copies.
The parity side needed nothing. gate_hooks and gate_events walk the live
catalog, so both seams were covered structurally as soon as the call sites
existed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Five small additions. None of them mention the mod that prompted them.
WorldAPI handles gain stepNow, canStep, placeAt, isMoving and setPassable.
scriptMove queues onto OverworldState.scriptMoves, and a non-empty
scriptMoves is how the overworld knows a cutscene is running, so it gates
handleInput -- an actor animated that way freezes the player's controls for
as long as it walks. Right for Oak marching to his lab, wrong for anything
moving on its own schedule. stepNow sets the same per-tile state without the
queue. It skips the collision check on purpose: the caller is replaying a
move that was decided somewhere else, and re-judging it here would let the
two copies disagree about where the actor is. canStep is there for callers
that do want the map's opinion.
OverworldController raises world.talk around the NPC talk path. An object
from spawnNpc has no TEXT_* id, so the vanilla path has nothing to say for
one; a mod that owns the object wraps this and does not call next.
LinkState.newFromSession adopts a transport that is already paired, so a
mode can tunnel a battle through a connection it already has rather than
opening a second one. The hello and fingerprint exchange still runs. When
the battle ends, link.battle_ended carries the result and both lockstep
party copies -- cable rules leave the real party alone, so that is the only
place the damage exists.
Game:startNewGame(opts) is the title's NEW GAME closure made callable, with
opts.intro=false to land straight in the world. CodeEntry.new takes an
optional {length=, charset=} so the same widget can enter an address or a
room code.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
bryanthaboi
github-actions
Shane McGovern
thibautbus
MaxTomahawk
1jamie
DESKTOP-8SRFDDM\cam95