Resolved where main's trainer meets the plugin branch:
- Board Break, the drone, Road Rash, Hall of Meat and Grind Exits are plugins here, so their
built-in code stays deleted: the trainer no longer hands Board Break's wear values back (the
plugin does not tune physics), its QUICK SETTINGS list no longer names Grind Exits or Board
Break, and the drone's debug action and menu are not brought back.
- The trainer's pump power, gravity, revert boost and transition help run beside the plugins'
pushes and velocities on each physics step; both sides' new declarations are kept.
- The game's own pad reads go through game_pad_read (the gamepad filter, Dark Pop's D-pad hiding,
the grind sticks) and then the plugin branch's menu and hidden-button handling.
- The trainer's forced grind settings are noted as the stance like any other value.
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.
TUNE as one page of folding sections, presets above the tabs, Skate 3 grind and
transition feel, Dark Pop, manual pop height, slow flips pop higher, per-trick
speeds, and a release audit of stock / resets / presets. See Extension/Trainer/CHANGELOG.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wk6Y7hSKoBANMm9ZHbStLg
Changing a setting in the game's menu did not stick. The camera went
back to high on leaving the menu, grind assist and lean exit stayed at
20 / on, the compass and scoring ticker kept their old state, and the
Experimental tab's "reset to default" returned to whatever was set at
launch.
b763631 gave the settings hooks per-thread caches of saved values that
look a value up again only when profile::Store::changes() advances.
Only whole-profile commits advanced it. Every single save (a setting,
an option, an outfit) goes through Store::Update::commit(), which did
not, so the caches answered with launch-time values for the rest of
the session. native_setting_get also wrote the cached value back over
the new one when the two differed. The saves themselves reached the
database, so profiles are intact.
- Update::commit() advances the counter through Store::note_change(),
after value_ holds the new values and before the lock is released.
Store::commit() and the constructor use it too.
- A cache hit needs the whole key text, not the key's address and first
8 bytes. Many keys share those bytes (egf_enable..., egf_exp_...,
CRAS_HasSeen...Tab), so a reused address could answer for another
setting. Still no heap string, store lock or map walk on a hit.
- An outfit that is not loaded or not saved is logged as a warning with
its reason and preset (local_cosmetic_loadout_rejected). It was a
debug event, so a player's log never said why an outfit was refused.
Routine cases stay at debug.
- dingosdk_profile_change_tests (with the park editor tests): each kind
of save advances changes(), an unchanged save does not, and saved
values keep their types across a restart.
Mesh streaming pool (fixes loading crashes on older GPUs)
- 1.0.0 forced MeshStreaming.PoolSize to 3.5 GiB on every card. An RX 580
crashed while loading with DXGI_ERROR_DEVICE_RESET, and the oversized pool
is the likely cause. The pool is now a quarter of the largest hardware
adapter's dedicated video memory, capped at 3.5 GiB; cards of 6 GB or less
keep the game's own 520 MB. The card, its memory and the chosen size are
logged once at startup.
Remote players (busy servers)
- Transform interpolate is inline plain arithmetic: 22 -> 5.8 ns per joint.
It was 4% of a busy server's client frame, since every player within
200 m has the whole skeleton interpolated every frame.
- Players out of the camera's view and more than 8 m away are sampled at
20 Hz, like far players. Their position still updates every frame for
collision, sound and nametags, and a player coming into view is sampled
that frame.
Expression hooks (run for every expression the game executes)
- No profiler zones in the expression hooks: taking a sample switched them
on, inflating the very samples that measured the hooks.
- The throwdown lab pump does one sorted lookup of the running graph and
skips everything else for graphs it doesn't watch (graph_throttle gets
throttled_graph for this).
Settings and memory reads
- Script gameplay-setting and native-setting gets are served from small
per-thread caches, keyed on the asset/key address, its first 8 name
bytes and a new profile::Store::changes() counter, instead of a store
lookup each call.
- Per-frame reads in the client tick, native settings and population
controls use memory::peek instead of ReadProcessMemory.
- Script error logging probes each argument pointer once.
Launcher
- "Close the launcher when Skate starts" now hides the launcher and exits
when Skate.exe does, instead of exiting as the splash opens. Outside a
Steam library, Skate's protection closed the game (TerminateProcess 0x32)
when the launcher exited during startup; players confirmed turning the
setting off avoided it.
- STATUS tile warns when Steam isn't open or signed in that PLAY starts
offline as Unknown Player (hidden when Offline mode is chosen or while
launching).
- Setting description updated to match.
Startup interventions (20260929)
- exit_return 0x87755f2 -> 0x87fe7fb: the last LEA RSP,[RSP]; JMP stub
before the PUSHFQ / LOCK CMPXCHG routine, as 0x87367cf was in 20260908,
and the return address player logs show. Only enforced once a frame site
is armed; this build's protector has no equivalent rep stosb frame clear.
Client frame cost (profiled 2026-10-02)
- Transform interpolate is inline plain arithmetic (~1% of an MP frame).
- Script gameplay-setting and native-setting gets are served from small
per-thread caches keyed on asset/key address, name prefix and a new
profile::Store::changes() counter, instead of a store lookup each call.
- Hot-path reads in client tick, native settings and population controls
use memory::peek instead of ReadProcessMemory.
- Script error logging probes each argument pointer once.
AyeZeeBB