A mod's reskate-plugin.json can declare "variants" {folder, order, displaces}: each <folder>/<name>/
is a variant, the one chosen (else the first of its order) loads, and every other mod whose layout
lists a displaced bundle is set aside for the run. Choices are kept in
%LOCALAPPDATA%\ReSkate\variants.json. Plugins read and choose their own mod's variants with
reskate.package.variants and reskate.package.choose (permission "variants"), and fit every board
for a deck that is not the game's own with reskate.board.fit {plates, axle}.
The board-break plugin, which ships inside the Board Break deck pack, declares the pack's decks as
variants, picks the deck on its own page and with `boardbreak deck`, reads each deck.ini and fits
the boards. Gone from the SDK: breakable_deck, the deck catalog, the MODS page's deck card, the
boarddeck command and the Board_Break_<Name> folder naming.
Plugins register modes (card, replaces, order, flag, players) and publish documents through
reskate.throwdowns.ui: the setup page, the waiting card, HUD slots themed on the game's HUD_<theme>_*
widgets, the match panel, flags and match restrictions. SmokeyDawg58's native plumbing
(throwdown_menu.cpp and its .inl files) renders them and posts throwdown_ui events back. kit::flag
is the one native flag registration. General services added for it: reskate.files, reskate.ui.dialog,
reskate.skater push/hold/bone_cam, reskate.camera.pad_flight, reskate.physics competition,
reskate.session.world, reskate.tricks feed/gesture, and world entities.
His ten modes are now Lua in his plugin: the rules, referee and wire (td/rules, referee, wire), and
every page and HUD document (td/ui*.lua). Extension/Modes and reskate.modes are gone, along with
every race and 1-UP branch in the SDK. The plugin test drives each mode through kit events.
A native setup page is never written to while it closes into flag placement: its rows are not
republished, the stock page is put back once the exit is over, and the stock Players slider is left
alone while a plugin flag is out. This fixes the crashes on the Deathrace button and on 1-UP's
Confirm.
reskate-plugin.json's "resident" list names the assets a mod's bundles keep (world objects and HUD
widgets the game picks at run time, and their shader block depots), so they survive another mod
replacing that bundle. Scanning now uses lexically_relative, so a mod folder that is a junction no
longer yields ../ paths, and every output path is checked to be inside the merge output. Chunk
metadata audit failures are fixed; the full merge audit passes.
reskate.throwdowns.mode{...} adds a mode to the game's own Throwdowns page: a card on
the picker, the real Spot Battle setup page relabelled with the mode's name, logo,
description, players, timer and rows (sliders, choices, buttons), and the native start
flag. The plugin hears throwdown_mode events (opened, setting, button, placed with the
flag's position and heading, start, leave, cancelled) and runs the mode itself;
release_flag takes the flag down. The flag's queue never starts a Spot Battle: its
waiting card's Start goes to the plugin, MaxPlayers is held at 10 and it is never
relayed as a drop. After SmokeyDawg58's one_up_menu and one_up_placement
(reskate-custom-mods), generalised over a mode registry.
The page is his 4-column logo grid whether or not a plugin has a mode. Its widgets and
grid records, and the game assets it uses that only some levels load, ship in
ReSkate.dll as the throwdown_menu built-in pack, taken from his asset mod's FBMOD
(dingosdk_ebx_pick trims the asset list in place). A mode's logo is a texture path in
its own mod, so no mod edits the shared list. The FPV drone is no longer a built-in pack.
Native menu data: text initialization, change-only writes, retained array publication,
owned-model rollback. Tests: registry and setup helpers, the plugin API, a test plugin
with two modes.
Hall of Meat (contrib/plugins/hall-of-meat) and grind exit triggers
(contrib/plugins/grind-exits) are Lua plugins: the bail model, its Meat card,
each map's best, the trigger rules, settings, console commands and menu panels
are theirs, with their own checks that dingosdk_plugin_tests runs.
What has to run in the game stays in core as plugin services:
- every physics step of the skater (skater_physics, the physics_step event,
permission events.physics);
- the skater's x-ray (xray, xray_render, xray_skeleton; reskate.skater.xray and
the draw handler's d:skeleton);
- skate.'s bottom left HUD corner (hud_corner; reskate.hud.corner);
- the game's slow motion, never in a session (game_speed; reskate.time.slow and
speed, permission game.speed);
- the side of the next pop out of a grind and pad controls kept from the game
(grind_exit; reskate.skater.grind_exit, the grind_exit event, reskate.input.hide);
- skate.'s own pictures by name (game_images; d:game_image), d:gradient and a
font for drawn text.
Removed: Extension/HallOfMeat, grind_exit_control and its commands, the HALL OF
MEAT and GRIND EXITS add-on pages, home tile, hub and native menu rows, their
menu models, and the hallofmeat and grindexit built-in commands.
The three features are Lua plugins in contrib/plugins (fpv-drone, board-break,
road-rash), built on what this adds to the plugin API: flight input, gamepad and
joystick reads with input capture, a rolled camera, a package's own sounds and a
synthesised motor (plugin_sound, plugin_wave, motor_synth), the skater's physics
events (skater_events), items worn over the outfit (outfit_overlay, skater_items),
the board's bones (board_pose), skater animation (anim_core), plugin images,
throwdowns started for chosen players, and server copies of plugins on a lobby
host and the dedicated server (server_plugins).
Removed from core: Extension/RoadRash, Extension/BoardBreak, the drone's flight,
radio, follow, OSD, beacon, outfit and sound code, its prop models (native_prop),
the DRONE, BOARD BREAK and ROAD RASH menu pages, home tiles and native menu rows,
their console commands, the drone follow debug action, the drone marker packet
(kind 36 stays unused; the roster keeps its byte) and the server's drones setting.
Kept in core, since every board needs it whichever plugins run: the breakable
deck packs (breakable_deck: which deck, its label and wheel spacing, the
boarddeck command and the BREAKABLE DECK card on the Mods page), and board_pose
keeping a deck pack's stretches out of sight, the baseplates on the deck and the
wheels at the deck's spacing on every board.
The console answered every command missing from ReSkate's own list itself, as unknown, so a
plugin's command (ptest) never reached the plugins on the game thread. Now only ReSkate's local
commands run there; anything else is queued, and the game thread tries the plugins first and
answers unknown as before when none has it. Checked in game: ptest all, 45 passed, 0 failed.
A mod with a reskate-plugin.json is a plugin: Luau scripts that run in their own VM, with only
the reskate API ReSkate registers for them. Luau 0.741 is vendored in External/luau (the VM and
compiler only: no native code generation, no loadstring, io or C modules).
- Limits: 64 MB of memory, 50 ms a call, 2 ms a tick on average (over it for 3 s stops it),
more than 5 errors in 10 s stops it, 50 log lines a second. A plugin over a limit cannot catch
it in a pcall and carry on.
- API: reskate.log, on (load, unload, tick, level_loaded, level_unloading), timers, the level,
its own package's files and JSON, require inside the package; console.command and print
("console"); storage get, set, add and keys ("storage"), 1 MB in
%LOCALAPPDATA%\ReSkate\Plugins\<mod>\storage.json.
- Found every 2 s among the mods mods.json leaves on: switching a mod, or changing a plugin's
files, loads, unloads or reloads it without a restart. A package holding a program is refused.
- plugin list, commands, reload, off and on in the console. A plugin cannot take a name of
ReSkate's own commands.
- Tests: DINGOSDK_BUILD_PLUGIN_TESTS, including plugins written to break the sandbox.
MY MODS: a table with an animated drag for the load order, a conflicts bar with Check conflicts, and a Resolve popup whose choices are kept. The check runs at startup and after changes; conflicts no longer show as problems, block Play, or hold mod files during an update.
Conflicts: only game EBX that cannot be combined, or Lua scripts with different versions; colourways and same-name additions are not conflicts.
Official mods: installed in Mods/<Namespace-Name> and turned on or off through mods.json, so they show in MY MODS; the old .reskate-official folder is moved over.
Merge: a scan cache (Mods/.reskate-scan.txt) keeps each unchanged mod's repair, mesh and texture scans, so big maps are not re-read every merge.
Launcher: an announcement from the backend shows under STATUS, with HIDE.
The device-removal watcher always runs, and the fatal handler notes a DXGI stop in gpu-breadcrumbs.txt. Crash reports carry a gpu.breadcrumbs tag (protocol 5).
Where two mods change one of the game's assets in ways that cannot both be
in the game (a resource or chunk changed differently, an EBX whose changes
cannot be combined, different assets added under one name), the one lower
in the load order is left out whole, so neither loads half its changes.
The merge used to keep the higher copy of each asset and lose the rest of
the lower mod's change silently, which broke Skaterize beside deckorate.
A check before the merge (find_conflicts) reads only the bundles two or
more mods ship and applies the merge's rule: about 2 s for 296 mods, on
several threads, each mod's copy of a bundle by itself. When it finds
conflicts nothing is built, the catalog leaves the losing mods out and
merges once. The merge keeps its own check behind it, tracking per asset
the last change a mod made itself, so a change carried through other mods'
copies is still caught. A conflict is remembered in .reskate-excluded.json
with the mod it lost to, and stays out while both are unchanged and the
winner still loads above it.
The launcher says "Checking the mods" and "Checking the mods for
conflicts" during those steps. Its mod problems screen says MOD CONFLICT
when that is all there is, and the Mod Manager shows a CONFLICT badge, an
N CONFLICTS banner, a Conflicts filter and, on the mod, which mod it lost
to and how to load it instead.
The launcher still reads the list at start and looks up each mod's latest
version, moves in a copy installed from Thunderstore and takes away retired
mods, but it downloads nothing by itself. Each card on the OFFICIAL tab says
NOT INSTALLED, INSTALLED, UPDATE AVAILABLE, DOWNLOADING or UPDATING, with an
INSTALL or UPDATE TO vX button (official_mods::install, on a thread of its
own), OPEN FOLDER once installed, and the On / Off switch beside its pills.
The banner counts how many are installed. Play waits for an install under
way.
Board Break and Road Rash start only when their official mod is installed
and on (mods::official_in_use), and say in the log which it is not; the
menus leave them out otherwise, as when turned off.
The live test (RESKATE_OFFICIAL_LIVE) checks that nothing downloads until
asked, then installs each.
The OFFICIAL tab has an on/off switch on each mod's card, with an OFF pill
while it is off. The list of those turned off is
Mods/.reskate-official/official-off.txt, one folder name a line, which the
launcher's sync never removes. A mod turned off stays downloaded and kept
up to date, but official_packs() leaves it out, so nothing merges it, no
deck comes from it and the in-game Mods page does not list it.
Board Break and Road Rash do not start while their official mod is off,
and say so in the log. The menu model carries that as `off`, and the side
rail, the addon tabs and tiles, the home tiles, the pause menu's settings
and the native tools menu leave them out without a gap. category_tabs
takes a list built at run time for that.
Tests: an official mod turned off stays in place and is left out, comes
back when turned on, and only a plain folder name is written.
The merge measures each custom map's own meshes, every level of detail of every MeshSet in its
level bundles, into Mods/.reskate/reskate-map-meshes.json. A custom map's mesh streaming pool is
the game's 520 MB and half as much again as those meshes, at most half the video memory free (as
before): most maps have 25 to 70 MB of meshes and reserved 1 to 2 GB for them. When the map needs
more than the card has room for, Mesh.GlobalLodScale goes from the quality preset's 1.25 or 1.5
to 1, so coarser levels take over nearer; the game's value comes back for the next map that fits.
A map's world textures (texture group Worlds_*) flagged to stream that keep every mip with the
bundle, as an older Studio built them (404 MB of PQS Cordoba's), are made to stream as the game's
own do: the bundle keeps the mip tail from a 64 KB block, both first-mip fields of the header and
the chunk record say so, and the sharper mips stream from the chunk the map's TOC lists in full.
Only in a map's own level bundles (Studio keeps cosmetics whole; changing those left a load at
Starting server), and only for chunks whose guid marks them compressed (the streamer reads any
other as raw pixels and copied past what it read).
Master Chief and TASM2 Spider-Man ship appearance presets whose EBX lists a type reference on the
slot of an import, which the game's load reads as a type index and crashes on (Skate.exe+0x3def285
and +0x3def262 in Sentry). The merge now checks every mod's own EBX, including bundles it would pass
through whole, and takes such entries (or a repeated one) out of the copy in the merged patch and of
the copies carried into other mods' bundles; the mod's original still feeds any combine, and the
merge notes say how many per mod and bundle. The check reads only the fixup lists.
A material lookup row pointing past the shader programs the game ships is left out of the merged
lookup, with a note, instead of reaching the shader registry. Tests for both.
The shader expression registry registers a database from a queued job, and builds its per-key item
lists from the database's lookup without a bounds check. A database destroyed before its job ran,
or a lookup index past its items, left the registry holding pointers that crash it later
(Skate.exe+0x3f7cf33, +0x3f7ceed, +0x20960b in Sentry). The guard skips a registration for a
database already destroyed and points an out-of-range index at a valid item of the same key,
logging each.
A rebuild builds one segment per key and each needed every key's slot, so a long replay built
them once per segment: the game's rebuild is hooked too and the slots built for its first segment
are kept for the rest of it (per Behaviour and Subject). The segment player is no longer hooked
(it only counted plays for keysmooth status, every frame). The smoothness is a slider beside the
switch in the overlay (sent when let go, as each change rebuilds the open replay's camera) and
Keyframe smoothness (100, 75, 50, 25%) in the pause menu and Mod Options > Visuals. The startup
line logs on the runtime channel.
A mod's item (or one of the game's own a mod changes) keeps three empty description lines, so
the game's item card, which grows upward from a fixed bottom edge, always has room in the same
place; the overlay draws the mod there: its icon (decoded off the render thread into a fourth
texture slot, the last 32 kept), a MOD or CHANGED BY MOD tag, its name, author, version and how
many items it has. The card is placed as the game places its own, measured at many window sizes:
by the smaller of its scales, a 21:9 frame centred on wider screens (the search bar follows it
too). The warning sign the game's focus script gives any description is taken off that notice
each tick while it shows the reserved lines.
The merge's item list records each mod's version and folder (an older list falls back to the mods
loaded at launch). The item grids now run every client tick instead of at the 500 ms
customization pace, follow the highlight between their 40 ms passes, and let go of a grid as soon
as its screen closes, so the card, the search bar and favorites keep up with the player.
The merge keeps the items the store-copy check reads in every mod and writes them to
Mods/.reskate/reskate-items.json (highest priority first, so an item two mods ship is the one
whose copy the game uses); a live merge keeps the entries of the mods it did not read again, and a
patch built before the file existed is built again once. The game reads it when it builds its
item catalogue: an item a mod adds says "From: <mod> by <author>" under its name, one of the
game's own that a mod changes "Changed by <mod>", unless it has a description of its own.
The game plays the replay camera's framing keys as cubic Hermite segments whose speed changes at
every key (the acceleration jumps there, and it brakes where a channel slows or turns back), so a
free-camera shot surges and settles around each key. keysmooth on rewrites each segment as the game
builds it, so that the game's own player plays the natural cubic spline through every key: speed
and acceleration carry on through the keys, and every key is still met exactly, at its time.
- keysmooth on|off (off by default), keysmooth amount 0..1 (1 by default; 0 is the game's curve)
and keysmooth status; kept in %LOCALAPPDATA%\ReSkate\KeySmooth.settings. Only the free camera
(Behaviour 1) is smoothed; the other camera modes keep the game's curve.
- A switch in Mod Options > Visuals and in the overlay's Skater > Freecam.
- The curve apart from the game, with tests (DINGOSDK_BUILD_REPLAY_KEYFRAMES_TESTS); the hooked
functions are fingerprinted, and a build that differs gets no hook.
- contrib/keyframe-smoothing.md: what it does, how, and what was seen in game.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A hitch on the render side never showed in a report: the client report lists slow client frames
only, and a present sample had no frames at all. While a present sample runs, the present hook
notes when each frame is presented; the report then lists the longest gaps between presents, each
with what the present thread's samples in it spent their time in and how many were waiting. This
is what found a 170 ms stall every 5 s: the game's batched GPU uploads blocking in the driver.
Scrolling shirts onto Yudi-Yuddispack's philippleinshirt_ap crashed the game (Skate.exe+0x3dc07ed):
one of the preset's set parameters has a type of 0x0c07010304040502 (null in another crash). The
game type-checks every set parameter when it next applies a preset to a material, reading the
type unchecked, and catching that crash inside addEsShaderPreset was too late: the preset was on
the material already and the next update crashed on it.
addEsShaderPreset's guard now checks the preset first, as that type check reads it: the TypeInfo,
its data, a word at +4. A preset with a parameter whose type cannot be read is left out as a
preset that is not loaded is (the item shows without it), and named in the log once with that
parameter's key. Many of the game's own types are built at run time, so the test is that the
type reads, not that it lies in the game's image.
The ReSkate menu now has two styles, switched with View / F11 or the buttons in its top bar
and saved with the profile (`ui fullscreen`):
- Window: the mouse menu as before, with a side rail of pages in groups.
- Full screen: made for a controller, in the pause menu's look. Pages across the top (LB / RB),
each page's sections as the ONLINE page's tabs (LT / RT), button prompts at the bottom, the
pause menu's sounds, and the pause menu's own Multiplayer pages as ONLINE.
The pages are regrouped: HOME (quick switches and a tile for each page), MAP, BUILD, SKATER
(freecam, first person, cheats), SANDBOX (once Trainer), ADDONS (drone, Board Break, Road
Rash, Hall of Meat, grind exits, music), ONLINE, PROGRESS, MODS, SETTINGS, DEVELOPER.
Full screen, every shared control is drawn for a pad: switches and option rows changed with
left / right one step a press, sliders as steppers that repeat when held, rows of buttons,
and pages that scroll with the right stick. Search boxes and typed values stay in the window.
A pad press takes the menu's focus back and hides the pointer until the mouse moves.
Also: map pictures in the map picker (official loading screens, a mod's icon) with a search
box in the window, Splash and Tutorial Island left out of it; number boxes beside the
window's sliders; the binds in groups, with an action bind to open the menu (`menu`); a menu
size for each style (`ui scalefull`); the time of day stepping from the one last asked for.
The launcher reads the official list from the ReSkate backend (or the last one it read, or its own)
and keeps each mod in Mods/.reskate-official at its latest version; a copy the player installed is
moved in, the deprecated Board Break retired. The game loads them always on and leaves out Mods
folder copies; the browser no longer lists them. The built-in zips are gone (FPV drone too), and
built-in folders no longer carried are removed.
Fourth Studio's Board Break pack (ten decks and every sound) and Road Rash wounds are built into
ReSkate.dll beside the FPV drone, unpacked under Mods/.reskate-builtin and always on.
- The Thunderstore copies (Fourth_Studio-Board_Break_Pack, Fourth_Studio-Road_Rash and the
deprecated Fourth_Studio-Board_Break, or the same put in by hand) are moved to the Recycle Bin
as the launcher starts and dropped from mods.json; the merge leaves them out should one still
be there, and the Thunderstore browser no longer lists, updates or installs them, a dependency
on one included.
- Board Break's deck list, sounds and deck settings look in the built-in packs too.
- Names with spaces inside them (Board Break's "Body Impact 1.wav") unpack.
- The build makes each zip in assets/builtin into one zstd stream (ReSkateBuiltinPacker,
Engine/Vfs/builtin_pack_format.h), which the DLL carries: repeated files across a pack cost next
to nothing. Board Break 15.0 -> 5.4 MB, the drone 6.0 -> 5.4 MB, Road Rash 36.1 -> 34.2 MB;
ReSkate.dll 86 -> 70 MB. Plain zips still unpack.
- Tests: the mods taken for built-in ones and their retirement, and a pack unpacking to exactly
its files, refusing one that leaves its folder or is cut short (RESKATE_BUILTIN_PACKS also
unpacks the packs a build made).
path::string() uses the ANSI code page and throws on a character it lacks (a Cyrillic folder on
an English Windows). The remaining places it was used now go through path_utf8:
- Mods: a deck folder or a mod folder named outside the code page no longer leaves the Board Break
decks empty, or the Mods folder ignored altogether.
- Game and mod archives: a file under Win32 with such a name is passed over instead of stopping
the read.
- pose-dump no longer ends the multiplayer session when the game folder is not ASCII.
- The item browser's status line and a Board Break warning no longer fail on such a path.
- Launcher: system fonts are named to ImGui in UTF-8, which it reads.
Sentry's minidump endpoint streams every attachment but the dump to an object store, which this
self-hosted Sentry does not run: ReSkate.log and report.json were dropped from every crash
(outcome upload_failed). Each crash now goes to the envelope endpoint as sentry-native sends one:
the event, the minidump, report.json and the log, all kept.
- The event carries the log's last 100 lines as breadcrumbs (debug lines only among the last 25),
the mods the game ran with (enabled, outdated, excluded), how long it had run, and an anonymous
install ID as its user. Its time is the crash's, however late it is sent.
- crash.site tags where it crashed ("Skate.exe+0x3d9ed9d"), as Sentry names game code by the
nearest export; mods.count how many mods were on.
- An accepted report is kept in sent/ (the newest 10) instead of deleted, so the player still has
their crashes.
- multipart.* is report_envelope.*; the receipt check also reads the envelope endpoint's
{"id": ...}.
combine() recorded every source's copy of every EBX and resource asset, so that conflicting edits could be combined later. Map mods carry near-complete copies of the game's root-level bundles, so most of those copies are the game's own bytes. A mod's copy equal to the game's first copy of the asset, the copy edits are combined onto, is no longer recorded; the game's copy, recorded first, stands for it. A contribution points at its source's folder instead of copying the path. The merge report counts the copies left out.
An array record's hash is one per field type in a document (every
array the base game's files hold follows this, and each such hash is a
key of the document's reflection group table, now read). A new array
takes the hash of an existing array of the same holding type and field;
callers can ask the writer to refuse when there is none, and the mod
merge keeps the old fallback. Checked on the game's own
level documents.
Each array record carries a hash the original files make specific to
the field or element type; the writer used to give every array the hash
of the first record of the same element type, so a rewrite changed
hundreds of them in bam's level root. Arrays read from a document now
keep their record's hash; only arrays with no source record borrow as
before.
A rewrite copied a boxed value's bytes, pointers and all, but wrote the
arrays and strings those pointers reach only when a field also reached
them. Rewritten documents lost that data and the boxed pointers aimed
at whatever moved into its place; the mod merge rewrites real documents
this way. Read documents now keep their EBXD bytes, and the writer
carries every array, string, instance and boxed value a boxed pointer
reaches, repointed. Found by a round-trip check on bam's level root (one string and
about 1,120 bytes of arrays lost).
The tests read the merged patch through CasStore and list_bundle, report a missing archive as a failure, and cover a mod archive numbered like a placeholder.
A mod archive numbered 65534 or 65535 is placed under a free index, so no real placement is read as a placeholder. carry() records every payload in one list (settle() still writes each once per install chunk) and keeps one archive index, as write() does.
The music mod's song names its audio chunk; each map that takes the song carries the chunk too, and every chunk entry in the patch points at bytes inside its archive.
A carried asset whose install chunk no package names fails when it is carried, and is noted, as it was before it was written once; it no longer fails the whole merge when it settles.
A carried asset was written again for every superbundle that carried it: 1.4 GB of 2.2 GB with 96 mods, which left no room for a live apply. Each one is now written once per install chunk, when its first superbundle settles.
A model is spawned at each other player's flying drone and moved every frame. It is the drone's
own model (FPV Drone by PinaPipe, CC BY 4.0, credited on the DRONE page), built by ReSkate Studio
into a pack that ReSkate.dll carries and unpacks into Mods/.reskate-builtin before the merge.
Without the pack the game's traffic cone stands in. dronemodeltest spawns one in front of you.
Most crashes are caught by the game's own fatal handlers, which write a
dump of thread stacks only; the launcher could only copy it after the
game had exited. A crash on a freed or overwritten object then showed
the faulting registers but not the object or the list it came from.
- game_fatal_dump: the game's MiniDumpWriteDump import points at
ReSkate. Before the game writes its dump, the crash reporter takes its
own of the still-running game for the faulting thread; the game's dump
follows as before. Reports say capture.source "game-fatal-handler".
- The reporter's dumps include the memory the stacks point at
(MiniDumpWithIndirectlyReferencedMemory), a few MB more a report.
- The shared report says who asked for the dump (protocol version 4).
- Tests: a game-handler crash is captured with a heap block only its
stack points at; ReSkate's own capture keeps it too.
A spatial transaction queue gives every object it tracks a cull ID from a
free list made once, when the queue is, at the game's
InitialCullIdListSize (40,000 in its config), and never grown. A custom
map with more objects (Cardiff) emptied it while loading and the next
insert read entry 0xffffffff of the list (crash at 0x489cfe9).
- spatial_cull_ids: as each queue makes its list, its size setting is
raised to 262,144, one ID for every entity slot the entity pages hold;
a larger configured value is kept. 12 bytes an ID per queue.
- The contract checks the list maker and how it reads the setting; it is
installed at startup with the other world pools.
719cba1 stopped writing the metadata pointers into the other renderer
(0x2daa00 bytes, Skate.exe+0x30c64cc, kept at 0x75f2700), taking them for
the state owner's alone. They are not: the other renderer's own code reads
them through the patched view (0x030e6bff, called from 0x30a7e18) and
append (0x03e94fe6, called from 0x30d8ebb). Without the write, it read
whatever lay 18 MB past its end. With it, as in every release before, that
memory was overwritten, or the write faulted when nothing was mapped there
(the store_field crash).
- The other renderer's allocation is patched to the state owner's
0x11ab080 bytes (0x030c64cc, mov ecx), so +0x11aae28/+0x11aae30 are
inside it.
- Both renderers' metadata pointers are written again, before and after
the native constructor. The resource-state and relation-mask vectors stay
the state owner's alone: the other renderer keeps fields of its own at
+0x2da920.
- The contract's comment says which renderer reads what, and where.
Players reported the game hanging for about a second while skating. A
profile showed the memory line in the log walking every heap
(HeapSummary) on the client thread, 0.5 to 2 s every two minutes. An
audit of everything that runs on the game's update and render threads
found more of the same, and Sentry's top crashes had causes of ours.
Freezes:
- Memory log line: no heap walk or thread snapshot; the GPU part and the
write happen on a thread of their own.
- Profile database: synchronous=NORMAL in WAL mode. A save is still kept
when the game crashes; every play event no longer waits for the disk
inside the game's hooks.
- ReSkate's own preferences and values (menu clicks, session sliders,
Discord) are saved by a background writer in order; reads see a value
until it is saved.
- Input: the foreground window and the keyboard are sampled on a thread
of their own (GetAncestor and GetAsyncKeyState stalled the client tick
40-50 ms when another program held Windows' lock); flight input uses
the sample; ClipCursor at most twice a second while a menu has the
pointer.
- Hooks installed while the game runs go in one transaction per place
(HookBatch): throwdowns, remote players, the party, Hall of Meat.
- Off the update thread: the host's map package lookup (one table read in
the background, not the Mods folder per offer), the trainer's map file
and settings save, Board Break's settings save, mod music playlists and
their covers, the Spotify station covers.
- Off the render thread: the Modding page's and the deck card's Mods
folder reads. The overlay's three startup waits share five seconds.
- Less work per frame: Steam names reused while one a frame is asked
again, 2 friends a tick, steam_api64.dll checked once per process,
cosmetic names read in place rather than a system call per character,
an outfit load no longer copies the inventory per slot, the throwdown
type scan runs at background priority over resident memory, the Board
Break deck label is kept a minute.
Crashes:
- Render-resource pool: the metadata fields are written to the state
owner only. The other renderer is 0x2daa00 bytes, and +0x11aae28 is far
past its end (store_field crash; heap corruption in every release).
- Other players' skateboard effects: a stale skateboard component is no
longer followed, and the joint lookup is walked with a bound first (an
endless loop in reused memory ended in GameStall).
- Launcher: Oodle decodes into the buffer it asks for with corrupt data
possible, and its assertions no longer stop the process (heap corruption
in the store-copy scan; a breakpoint in oo2core).
Timings (Send performance timings):
- A frame of half a second or more in a level is sent as game.freeze:
where the time went (ReSkate's tick, the game's frame, or outside it),
whether the renderer kept presenting, and ReSkate's busiest zones.
- Closing the game sends the level session it was in.
Music after a connection drop:
- Spotify device: an account Spotify could not be reached with is kept and
tried again (3 s doubling to a minute); a session that drops is taken
up again. Only an account Spotify refuses is asked for anew.
- YouTube playlist pictures that could not be fetched are tried again,
from 30 s to 10 minutes apart.
Switching time of day could load the new layer while the old one was
still loaded: the apply loop goes through the layers in catalog order
and sends one event a pass, so going to an earlier time (WeatherDay ->
Evening) sent the load first, and nothing waited for an unload to finish.
Two times of day at once carry both times' lights, and a level with many
lights overflowed the engine's light tables.
A time-of-day layer is now asked to load only once every sibling time of
day that is to be off reports requested and loaded both false; until then
its row says "Waiting for the old time of day".
An unload is no longer sent while the engine is still loading the layer
itself. The engine ignored it, so at a level load the level's own time of
day could stay loaded beside the saved one, and with the wait above the
saved one never loaded.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
AyeZeeBB