UgcToolbox runs LU Toolbox itself in an external Blender that stays up
and makes one model after another (dlu_toolbox_worker.py runs
LU-Toolbox-Standalone's steps: import, Process Model, Bake Lighting,
niftools export). JSON lines over its stdin and original stdout; it is
restarted after a crash, a timeout, 100 models or new settings, runs at
worker_nice with toolbox_threads, and its CPU time is charged to the
worker, which pauses it (SIGSTOP) while the CPU budget is overdrawn.
ProcessModelToolbox reads LU Toolbox's .nif back, counts its triangles
and draws the icon from it. Tests: option parsing, the fallback, the
protocol framing, the worker with a stand-in Blender (crash, hang,
failure, give-up), and LU Toolbox itself when DLU_TEST_TOOLBOX_* are set.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
After serviceType the reply now sends major, minor, patch, a flags byte
(build kind in bits 0-1, dirty in bit 2), the first 32 bits of the commit
hash and a u16 length-prefixed build string, instead of the stale fixed
ASCII "0.1.3". unknown stays "DLU3". The 1.10.64 client reads only
netVersion and serviceType and never checks the length, so the extra
bytes are ignored. The build string is optional on read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Live GF Large Crates (LOT 1859, path CrateMast) and FV small white shrines
(LOT 3141, path gate_statue_quickbuild) dropped 1-point powerups outside
their template matrices, from the path's smashable_loot_matrix=1:29 with
smashable_loot_matrix_set=7:1. DLU already carries a spawner path's
waypoint config into the spawned entity's settings; this pins it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A quickbuild's HasItem preconditions took their items only when the build
completed and never gave them back. Live took them when the build started
and gave them back when it was cancelled: the FV Stone Warrior pedestal
(LOT 8551, precondition 99: 5 of LOT 6194) took the items at Building 5
times and added them back with the loot source Quickbuild on each of the 3
cancels in the live captures.
Preconditions now report their item costs instead of removing items while
checking. A quickbuild takes them at the start, gives them back on cancel
or a reset during the build, keeps them on completion, and gives them back
when the builder leaves the world mid-build.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ActivityRewards coins were always read from CurrencyTable npcminlevel 1.
Live used the reward row's ChallengeRating as the npcminlevel, and level 1
when the currency index has no row for that level:
- FV foot races (ChallengeRating 4, index 1) gave 36 and 48 coins, which
only level 4 (30-50) fits; level 1 is 3-5.
- Frakjaw's chest (activity 58, ChallengeRating 6) gave 250 each to a team
of 2: its currency indices 123-126 only have a level 6 row (500), so DLU
gave nothing.
- Quickbuilds, wishing wells and chests have ChallengeRating 1; survival
and the shooting galleries have ratings with no row and keep level 1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The third ray backend, so every machine has a library for it: Embree on x86
CPUs (embree), HIPRT on AMD and NVIDIA GPUs (hiprt), Embree through SYCL on
Intel Arc and Xe GPUs (embree-gpu).
- CMake option DLU_EMBREE_SYCL (off). dUgcServer/EmbreeSycl is a project of
its own built by a SYCL compiler (DLU_SYCL_CXX; icpx through ONEAPI_ROOT or
the path, or the open source DPC++'s clang++ through DPCPP_ROOT) as an
external project: Embree 4.4 with EMBREE_SYCL_SUPPORT, linked statically and
bound inside (-Bsymbolic, only its C functions exported, so it never meets
the servers' own Embree), and the GPU kernels (nearest hit skipping a ray's
triangle, any hit), into libdlu_embree_sycl next to the servers
- UgcRaysEmbreeGpu loads it the first time embree-gpu is asked for; one GPU for
the process (embree_gpu_device picks it), the workers take turns, the
occlusion rays in batches as for hiprt
- without the build, the library or a supported Intel GPU it falls back to
embree and says why (the UGC server's log at start, --make-model on stderr)
- the option names, the settings page, the dashboard's picker, /reprocessproperty
Verified here: the default build and ctest; the SYCL build with the open source
DPC++ 7.1.0 (compiles, links against oneAPI's libsycl.so.9, exports only its C
functions); on this machine (no Intel GPU) it loads, finds no GPU and falls
back to embree. Not verified: tracing on an Intel GPU (none here).
Check: on a machine with an Intel Arc or Xe GPU and oneAPI, configure with
-DDLU_EMBREE_SYCL=ON and run UgcServer --make-model x.lxfml out embree-gpu;
the UGC tests then compare it with Embree.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Embree 4 (always built) replaces the UGC server's own bounding volume
hierarchies (the nearest-hit one and the occlusion rays' any-hit one), with no
fallback to them. ray_backend is embree (default) or hiprt (optional build;
Embree when it can't be used).
Settings, stored options and stats that say builtin still read: it is embree
(UgcRays::Parse, UgcProcessOptions::Parse). The dashboard's picker,
/reprocessproperty and --make-model offer embree and hiprt.
Tests: the backends are compared with Embree (hiprt when built), Embree against
rays whose hits are known, and the clutter's occlusion against what the old
hierarchy worked out (296 vertices summing to 114.5, 26 open, 183 dark); the
pinned model hashes are unchanged with Embree.
Check: ray_backend=builtin in an ini still starts and uses embree; the
settings page offers Embree and HIPRT.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The path tracer that imitated LU Toolbox's Remove Hidden Faces took 17 to 100
times as long as the renders from 42 directions around the model, too heavy to
use. It goes with its settings (hsr_method, hsr_samples, hsr_bounces,
hsr_sample_spacing, hsr_min_points), its side by side tracing for GPUs and its
tests. The renders (UgcRender::VisibleFromAround) remove hidden faces as before
that method existed; their size is hsr_resolution (1024), and the memory
estimate counts their buffers again.
The hidden-face method is no longer a processing option: the dashboard's picker,
/reprocessproperty and --make-model take only the ray backend (the occlusion's)
and denoising. Options stored before (made_options, process_options,
ugc_process_runs, --make-model arguments) that name toolbox or fast still
parse; the word is skipped.
Check: the settings page has no hsr_method or path settings and has
hsr_resolution; /reprocessproperty embree toolbox still works (toolbox
ignored); models made with the defaults keep their hashes (UGC tests).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
HIPRT (MIT) behind the CMake option DLU_HIPRT (off): its headers come from its
SDK (HIPRT_ROOT, else ROCm's /opt/rocm) and are copied next to the servers; its
library is loaded when first used (hiprtew), as HIP or CUDA are by Orochi
(MIT, fetched pinned by hash; CUDA when its toolkit is found). The trace
kernels (nearest hit skipping the triangle a ray leaves, any hit) are compiled
the first time and kept in cache/hiprt. One GPU context for the process
(hiprt_device picks the GPU); the workers take turns on it. When HIPRT, the
GPU or a scene's upload fails, Embree is used instead, and the UGC server logs
why at start.
For a GPU the rays go in batches (UgcRays::Scene gets batch queries; the CPU
backends answer them a ray at a time):
- hidden faces: with a batch backend the paths are traced side by side, a
bounce at a time (the path code split into Start, Scatter and Bounce, the one
by one tracing unchanged); the same paths with the same random numbers, so
the same triangles are decided (tested with builtin side by side)
- the occlusion bake and the denoised icons' traced occlusion always ask in
batches (the same rays, the same results)
Its symbols are hidden: the servers export theirs (-rdynamic), and HIPRT's
library, which has an Orochi of its own, would otherwise call ours.
Check: configure with -DDLU_HIPRT=ON on a machine with ROCm (or HIPRT's SDK)
and an AMD RDNA or NVIDIA GPU; UgcServer --make-model x.lxfml out hiprt; the
UGC tests (hits, hidden faces and occlusion against builtin); a build without
it leaves everything as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Intel Open Image Denoise 2 (Apache-2.0) behind the CMake option DLU_OIDN (off):
an installed OIDN is used when found, else Intel's release package (pinned by
hash) is downloaded and its libraries copied next to the servers.
A denoiser only removes noise that differs from pixel to pixel; the icons'
occlusion comes from the bake, per vertex, which it leaves as it is (checked:
white noise 0.17 -> 0.006 relative spread, per-vertex blocks unchanged). So
with denoise=oidn a model's icon is drawn from model.noao.nif (its colors
before the bake) with its occlusion traced per pixel of the supersampled image
(denoise_samples rays, default 4, with the bake's distance and strength and the
ray backend), box filtered and denoised at the icon's size, guided by the colors
and normals. The model keeps its baked occlusion. Icons drawn again from stored
files use the stored model.noao.nif the same way.
OIDN works on a thread of its own; its time is charged to the job's thread
(UgcThrottle::Charge), so the CPU budget and the recorded CPU time include it.
Check: configure with -DDLU_OIDN=ON; UgcServer --make-model x.lxfml out oidn
and compare its icon.png with one made with off; an OFF build leaves icons as
they were.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Staff can make models again with other processing options than the UGC
settings' (ray backend, hidden-face method, denoising) and every make records
what made it, for comparing the options.
- migration mysql 99 / sqlite 82 (ugc_process_options): ugc.process_options
(picked for the next make, cleared once made), ugc.made_options (what made
the current files) and ugc_process_runs (every successful make: options,
wall and CPU time, hidden faces', occlusion's and icon's time, bricks,
triangles before and after)
- IUgc: ResetUgcModelProcessing and ResetPropertyUgcModelProcessing take the
options; PendingModel carries them; RecordUgcModelRun, GetUgcRunSummaries;
list entries have madeOptions and processOptions
- the UGC server applies a job's options over its settings and records the run
- /api/ugc/reprocess takes options ("embree fast oidn"); /api/ugc/options
lists the choices, the settings' defaults and averages per combination
- /reprocessproperty [builtin|embree|hiprt] [toolbox|fast] [off|oidn], any
order, all optional
Check: run the migration on MySQL and SQLite; /reprocessproperty embree fast
on a property, then the models' made_options and ugc_process_runs rows;
/api/ugc/options.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New UGC settings (ugcconfig.ini and the dashboard's settings page):
- ray_backend: builtin (default), embree or hiprt (falls back to embree when
the build or machine can't); the hidden faces' paths and the occlusion rays
- hsr_method: toolbox (default, LU Toolbox's paths) or fast (the renders from
around the model), and hsr_fast_resolution (1024) for the fast one
- denoise: off (default) or oidn (icons; off until a build has it)
UgcProcessOptions (dCommon/UgcKeys.h) names the choices for everything that
passes them on ("embree fast oidn", any order, left out: the setting's).
UgcJobs::ApplyOptions puts a choice over the settings and MadeWith says what a
make used after fallbacks; a made model's stats.json records it (rays,
hsrMethod, denoise). UgcServer --make-model and --make-modular take the
choices after the folder and print the CPU time and what made it.
Check: the defaults make the same files as before; the settings page shows the
four settings under UGC; UgcServer --make-model model.lxfml out embree fast.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
UgcHsr::Options::method picks how hidden faces are found: toolbox (LU Toolbox's
paths, the default, unchanged) or fast, the test the UGC server used before
"hidden faces removed as LU Toolbox's Remove Hidden Faces decides them",
restored unchanged as UgcRender::VisibleFromAround: the opaque mesh rendered
from 42 directions around the model (Options::fastResolution pixels square,
1024 as before) and the triangles that show in none removed. It is much
faster, but it also removes faces seen only by bounced light (interiors,
recesses), which LU Toolbox keeps. Nothing sets it yet.
Check: the UGC tests (the fast method removes a box seen only through a
chimney, which the paths keep; its files are the same every time).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
UgcRays::Scene answers the two ray queries the UGC server makes: the nearest
hit for the hidden faces' paths (never the triangle a path leaves) and any hit
for the ambient occlusion rays. Two backends:
- builtin: the two hierarchies the queries had before, moved unchanged (each
built the first time it is asked), so the files made are the same bytes
- embree: Embree 4 on the job's thread (a device per worker thread, no threads
of its own, so its time counts in the CPU budget), watertight, the skipped
triangle filtered out
UgcHsr::Options::rays and UgcRender::AoOptions::rays pick the backend; both
default to builtin, and nothing sets them yet.
Check: the UGC tests (builtin's files keep their hashes; UgcRays tests compare
embree's hits, hidden faces and occlusion with builtin's).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Item::RemoveFromInventory ended with `delete this`, so every caller that still used the item afterwards (SetCount(0)
returning into its caller, loops that remove several items, proxies purged while their parent is handled) touched
freed memory. The inventory now takes the removed item and frees it at the inventory component's next update, or
with the inventory.
Check in game: sell, drop, delete, trade, mail and use up stacks (including the last of a stack); unequip and
remove an item set piece and a proxy-bearing item (rocket, modular car); donate items; nothing crashes and the
inventory shows the right counts after relogging.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The client's TacArcBehavior::DoHit (0x00fb10c0) writes the closest max-targets ids from a set, so ascending and
each once, then the action data per id in that same order; DoUnserializeBS (0x00fb26a0) reads them the same way and
skips empty ids. The server handled the targets in list order and skipped ids whose object it could not find
without reading their action data, so every later target read the wrong bits (several pirates under a Doom Slicer).
Handle now reads the action for every listed id in ascending order, and the server's own casts write ids and actions
in ascending order after picking the closest targets.
TacArcBehavior::Cast (0x00fb2d10): a picked target that passes the filter gets the action with no TacArc data (the
server's own cast now calculates that action instead of handling it); otherwise the target is dropped, and an arc
measured from the target's position writes nothing.
Check in game: Doom Slicer and multi-target katanas on groups of pirates/admirals damage each of them; apes still
take damage during their stun; enemies with arc attacks (apes, Maelstrom horsemen) still hit players.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The live boss script's rain of fire takes every target of the first ROF target group and ROFImpactCnt (2) different
random targets of each other group. The server took one per group, so the rain was sparser than live. The impacts are
now picked like the live script (without repeats inside a group).
Check in game: AG Spider Queen stage 3; the rain of fire lands on the centre ring and on two spots in each outer ring.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The live boss script tracks which arena zone volume (Zone1Vol..Zone8Vol, TeleVol for the default Zone3Vol) each
player last entered, and its rapid fire picks a random player, takes the three RFS target groups around that zone,
sorts each by the targets' CWOrder (CWOrder2 when the sweep crosses between zones 8 and 1), clockwise or
counter-clockwise at random, drops the first and last target of the middle group (shared with its neighbours),
turns with skill 1480 at the fourth target and fires 1394 at every target in turn, playing attack-shoot-right or
attack-shoot-left. The server shot a single random target with the single-shot animation, and the AG property zone
never subscribed the boss to the zone volumes. The zone now registers the volumes (retrying until they are spawned)
and the boss builds the sweep like the live script.
Check in game: AG Spider Queen stage 2; the rapid fire is an arc of many shots sweeping across the arena near the
player, left or right, and follows the player to other parts of the arena; after teleporting it starts from the
default side.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every hit started both the rapid fire shooter and the rain of fire timers, from stage 1 on, so the two specials
ran in every stage and on top of each other, each turning the boss's AI off and on again under the other's
animations. They also wrote the "stoppedFlag" that the no-players-around attack stop uses, which could leave her
stopped for good. As the live script: after she comes back down, a skill manager fires the rapid fire shooter in
stage 2 and the rain of fire in stage 3 only, again 10 to 15 s after each ends; for 3.1 s after her melee smash
(skill 322) a due special waits and fires when the smash ends. The rain of fire keeps her from attacking until
its last impact.
Check in game: Spider Queen fight: no specials before the first spiderling wave; stage 2 only rapid fire, stage 3
only rain of fire; she never freezes in the smash animation and keeps attacking after each special.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The zone script never handed the boss her landing target and scream emitter: ZoneAgProperty::ProcessGroupObjects
was empty and nothing answered the boss client script's "QueryZoneScript" event. As the live scripts: the boss asks
the zone ("RetrieveZoneData"), the zone stores the first object of Land_Target and Spider_Scream on her as
LandingTarget and ScreamEmitter (looking again every 0.3 s until they are spawned), and each spiderling death sends
NotifyClientObject "EmitScream" with the emitter, which the boss's client script plays as the scream. The landing
skill and camera shake now come from the landing target, not the boss.
Check in game: AG Spider Queen (property or instance): kill a spiderling: the scream plays from the mountain; when
she comes back down, the landing blast hits around the landing spot and the camera shakes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The zone file tests' sample zones now use generic world IDs, file names
and path names instead of ones taken from real files. What they test is
unchanged.
Check in game: nothing (test only).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
LDF_FROM_STRING (1.10.64, 0x010fa220) splits an object's config at
every comma and line break, leaves out empty entries and entries with
no '=', and sets each key=value it finds. The reader split at line
breaks only and parsed every piece, so an object with no config got an
empty entry of unknown type, and a value with a comma would have stayed
whole.
Live scene files are unchanged; objects in older client files with an
empty config no longer get the empty entry.
Check in game: every world spawns its objects as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ClientZoneFilesTests now
- reads sd0 files instead of skipping them: some clients' loose sd0
files hold what followed the file in its pack, and the file is the
first sd0 stream, up to the first chunk that does not fit;
- reads the terrain (.raw) of every zone of version 30 or newer, once
per terrain file shipped by several clients;
- skips a damaged file (by the MD5 of its bytes) only when it does not
read, so only the files the client cannot read either are left out:
a scene file that claims 81 objects and ends after the first, and six
version 30 terrain files with a width x width scene map per chunk
where the client reads one byte (RAWReadSceneMap, 1.10.64) and so
misreads every chunk after the first.
Every other zone, scene and terrain file of every client on disk reads,
in about 80 seconds.
Check in game: nothing (test only).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The client reads a terrain chunk of a file older than version 32
(1.10.64):
- its color map as width x width BGRA pixels, keeping the
(width - 1) x (width - 1) before the last row and column, as RGBA
(RAWReadColorandLightMaps); the reader kept all the pixels as
written;
- its texture blend map's pixels as BGRA, kept as RGBA (0x0103aaf0);
- before version 31 no scene map, only a byte: the client's scene map is
all scene 0 (RAWReadSceneMap); the reader had none.
A chunk of width or height 0, which the client reads (no heights, no
color map), no longer fails the whole file.
Live terrain files are version 32, so they read as before.
Check in game: nothing to check (no live terrain changes).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ReadLvlObjectData (1.10.64, 0x0103ba20) changes an object's config as it
loads it:
- before version 47 a respawn time is in milliseconds: a float over 100
or any u32 becomes that many seconds, as a float;
- a model spawner (LOT 176 with spawntemplate 14) gets its subkey as its
blueprintid unless it has one, DisableModelBehaviors true unless set,
and preventRenderWrapping true;
- before version 44 an object with sceneIDOverrideEnabled gets
sceneLayerIDOverride 0.
LevelFile now makes the same changes. On live worlds 1906 spawners in
114 scene files of versions 45-46 have a u32 respawn time; the world
already divided those by 1000 but as an integer, so they now keep their
fraction (a 2500 ms respawn is 2.5 s, not 2 s). No live scene has a
model spawner; 62 objects of version 43 scenes get sceneLayerIDOverride.
Check in game: enemies and smashables respawn after the same time as
before on Forbidden Valley, Avant Gardens and Gnarled Forest.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SceneLoader::ReadLvlFile (1.10.64):
- chunked files: the file info chunk starts the file and gives where
the environment, object and particle chunks start, 0 for none
(DoLvlChunk, ReadLvlChunk1000); the reader walked the chunks one after
another by their sizes instead;
- older files: the editor settings are a u32 size and that many bytes
(SceneLoader::ReadEditorSettings, skipped by size); the reader took
them for a u32 and then a count of 12 byte points;
- older files before version 3 have no objects for the client ("Level
file is unsupported").
LevelFile now does the same. Every scene file on disk reads the same
objects as before.
Check in game: every world spawns its objects as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ReadLvlObjectData (1.10.64, 0x0103ba20) reads a scene object's last
field, from version 7 on, as a render technique count, and when it is
not 0 a 64 byte block and 133 bytes per technique (a 64 byte name, a
u32, a u8 and 16 floats) follow. The reader took the count for an
unknown u32 and read the next object from inside the techniques, and
read the u32 for every version. It also left the node type and glom ID
uninitialized when a file has none (or, for the node type, one outside
0-10); the client uses 1 for both.
The count is now SceneObject::renderTechniqueCount, the techniques are
skipped, and the defaults are the client's. Every scene file on disk
has no render techniques, so what they read is unchanged.
Check in game: every world spawns its objects as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
LevelPath::FromBuffer (1.10.64) changes two values as it reads a path:
- a property path's reputation multiplier outside 0-1000 becomes 0;
- a camera waypoint's FOV in a path older than version 11 is divided
by 1.25.
The reader kept both as written. Only camera paths of older client
files (versions 8-10) change; no live zone has either case.
Check in game: nothing to check (no live zone changes).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The client reads a spawner path's fields (LOT, respawn time, counts,
object ID) and a camera path's next path and rotate-player flag only
when the path is newer than version 3 (LevelPath::FromBuffer, 1.10.64:
pathVersion > 3). The reader read them for every version, which would
misread an older spawner or camera path from there on.
No zone file on disk has a spawner or camera path older than version 4,
so what they read is unchanged.
Check in game: nothing to check (no live zone changes).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The client reads a zone file's paths as a chunk of its own
(LuzFile::ReadLUZFile copies the u32-sized chunk, ReadLUZPaths reads
it, 1.10.64) and drops every path when it refuses any of it:
- a chunk version of 2 or more, or 10000 paths or more (ReadLUZPaths);
- a path of version 19 or more, 10000 waypoints or more, or a
waypoint with more than 99 name/value pairs (LevelPath::FromBuffer).
The reader read the paths straight from the file, trusting the chunk to
be well formed. It now reads the chunk by its size, reads the paths
from it with the client's limits, and when one is refused the zone has
no paths (logged), while the rest of the zone file still reads.
Every zone file on disk reads the same paths as before.
The unit tests' sample zones now give the path chunk its real length.
Check in game: moving platforms, NPC patrols and spawners work as
before on every world.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Scene transitions of versions 34-38 have five points. The client links
the scenes of the first and the last one (ZoneLoader::ReadZoneFile,
1.10.64: points[0] and points[count - 1], for every version); the scene
graph used the first two, so in those zones it linked a scene with the
one of the transition's second point instead.
Of the 130 five-point transitions on disk, six (all in one older client
zone) have a different scene at the second and last points; the rest
link the same scenes as before.
Live zones are unchanged.
Check in game: nothing to check (no live zone changes).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
LuzReader::ReadScenes (1.10.64) keeps a zone's scenes by scene ID and
layer, so a later scene with the same ID and layer replaces an earlier
one. ZoneLoader::ReadZoneFile then goes through them in scene ID order
and
- leaves out every layer of a scene ID that has no layer 0 (General)
scene;
- before version 41 loads only the layer 0 scene of each ID; from 41 on
the other layers (Audio, FX) are loaded with it.
ZoneFile now keeps the scenes the same way, so the world loads the same
scenes as the client. No zone file on disk has a duplicate, a scene
without a General layer, or a non-General layer before version 41, and
their scenes are already in ID order: what every file reads is
unchanged.
Check in game: nothing to check (no live zone changes).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Zone files before version 33 have no scene IDs. The client gives each
scene its index in the file as its ID, -1 past 255, with layer 0
(LuzReader::ReadScenes, 1.10.64, via the LWOSceneID from the loop
index); the reader left every such scene at ID 0, so a zone with several
scenes had them all collide on one ID.
The version 30 zones on disk have one scene each, so what they read is
unchanged.
Check in game: nothing to check on live worlds (all are version 36+).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Zone files older than version 30 were refused. The client reads them
(1.10.64):
- LuzFile::ReadLUZFile reads a version below 20 as 20;
- LuzReader::ReadScenes reads a scene of a version 20-29 file as a bare
u32 SceneTable ID, with no file name;
- ZoneLoader::ReadZoneFile takes those scenes in SceneTable ID order,
looks each one up in the CDClient SceneTable and, when there is a
row, makes it a scene with the next index (0, 1, ...; -1 past 255) as
its ID and the row's sceneName as its file; a scene with no row is
left out.
ZoneFile now reads these files (FileFormatVersion::Oldest = 20 for
anything older), keeps the SceneTable ID on ZoneScene::sceneTableID,
and ZoneFile::ResolveSceneTable does the lookup the client does; the
world looks the names up in the CDClient SceneTable.
The only such file on disk (version 20) now reads.
Check in game: nothing to check on live worlds (all are version 36+).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The world side of guilds (docs/Guilds.md):
- TMP_GUILD_CREATE, which the client's guild create box sends to its world, goes to the chat server as GUILD_CREATE.
- The character component reads the character's guild from the database when it loads (instead of the old unused "gn"
and "gid" charxml attributes) and takes GUILD_GET_STATUS from the chat server; a change is serialized once (the
client redraws the name billboard every time it reads a guild name). A name waiting for moderation isn't shown.
- Slash commands (the client has none for guilds): /g and /guild (guild chat; the client's guild tab sends /g, which
goes through the chat filter and mute like zone chat, then to the chat server as channel 10), /guildcreate (opens the
create box with DisplayGuildCreateBox), /gkick, /grank <name> <officer|veteran|recruit>, /gleader and
/gdisband confirm.
- The Guild Master script (LOT 3001, L_GUILD_CREATE.lua: using it opens the create box). Live never placed it; GMs
can spawn it.
Check in game (two accounts, a client with a FeatureGating row "guilds", 1, 0, 0 in its cdclient.fdb): /guildcreate,
make a guild (the name shows under yours, the guild button appears on the status bar); invite the second character
from the guild window, accept, both lists show both; guild chat tab; /grank, /gkick, /gleader; leave from the window;
log the second character out and in (guildmate logged off/in lines).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The chat server is the guild authority (docs/Guilds.md). GuildManager keeps guilds in the database (IGuilds) and tells
clients through their worlds, with every outside dependency behind hooks so the rules are tested without a server:
- Create: name rules (3-30 of letters, digits, space ' - ., no space at either end or twice), the chat filter's deny
list refuses a name (BAD_NAME), names not on the allow list make the guild but wait for moderation (other players see
no guild name until then), a taken name (case ignored) is EXISTS, someone in a guild can't make one.
- Invite / answer: leaders and officers invite; the client's answers for not online, already in a guild, invite pending
and could not invite; one invite per player, answerable for 10 minutes and gone when the player logs off; a full guild
(guild_max_members) takes nobody. The new member is a recruit; the inviter hears the answer and gets the list again,
the other members get GUILD_ADD_PLAYER.
- Leave, and DLU's kick, rank and disband (the client has no controls for them): the leader kicks anyone, officers kick
veterans and recruits; a leader who leaves hands the guild to the highest-ranked, longest-serving member, and the last
one out ends it. Rank changes send everyone the list again (the client's GuildSetPlayerRank does nothing).
- GUILD_DATA for GUILD_GET_ALL, login/logout/world-change updates to guildmates, a guild that lost its leader (character
deleted) gets one again, and GuildChanged catches online players up with what the dashboard did.
- Every change is a guild_events row. Each character's world gets GUILD_GET_STATUS with the guild and the name others
may see.
Not connected to the chat server's packets yet.
Check: GuildManagerTests (18 tests) run every rule against an in-memory IGuilds. Nothing to check in game yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The guild packets of the 1.10.64 client (docs/Guilds.md) as Serialize/Deserialize structs, laid out the way the client's
packet handlers read them: GUILD_CREATE_RESPONSE, GUILD_INVITE, GUILD_INVITE_INITIAL_RESPONSE, _FINAL_RESPONSE,
_CONFIRM, GUILD_ADD_PLAYER, GUILD_REMOVE_PLAYER, GUILD_LOGIN_LOGOUT and GUILD_DATA (ClientPackets), the world packet
TMP_GUILD_CREATE the create box sends, the chat packets the client sends through its world (GUILD_INVITE,
GUILD_INVITE_RESPONSE, GUILD_LEAVE, GUILD_GET_ALL), DLU's world <-> chat packets (GUILD_CREATE, GUILD_KICK,
GUILD_GET_STATUS as the chat server's guild update to a world, and GUILD_SET_RANK and GUILD_DISBAND appended to
MessageType::Chat after CREATE_TEAM) and the game message DisplayGuildCreateBox (626). Fixed-size names always keep a
NUL, since the client reads them as C strings. Enums eGuildCreateResponse, eGuildInviteResponse,
eGuildInviteFinalResponse, eGuildRank and eGuildLeaveReason. Nothing sends them yet.
Check: GuildPacketsTests compares every packet with the bytes at the client's offsets. Nothing to check in game.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Four tables (MySQL migration 97, SQLite 80) and the IGuilds interface with MySQL, SQLite and TestSQL implementations,
for the guild system (docs/Guilds.md). A guild has a name (unique without regard to case), a name status like
pet_names (1 waiting for moderation, 2 approved), its founder and when it was made. A character is in at most one guild
(guild_members, with its rank: 1 leader, 2 officer, 3 veteran, 4 recruit, the client's names, and when it joined) and
has at most one pending invite (guild_invites). guild_events is each guild's history and outlives the guild. Deleting a
character removes its membership and the invites to and from it. Nothing uses the tables yet.
Check: both migrations run on a fresh and an existing database; DatabaseParityTests Guilds passes against MariaDB
(DLU_TEST_MYSQL_HOST) and SQLite gives the same results. Nothing to check in game.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A make's recorded time (process_ms, stats.json's ms.total) includes the
icon drawn at its end. Drawing only the icons again (the icon editor's
Draw all icons of this type again) left the stats and the rows with the
old icons' time. Now the icon-only job writes stats.json with the new
icon's time (ms.icon, ms.total changed by the difference,
UgcJobs::WithIconTime) and the main thread changes the row's process_ms
and process_cpu_ms by the same difference (the icon is drawn on one
thread, so its time is taken as its CPU time). The dashboard's Took,
CPU and totals follow.
Check: open a player model on the UGC page, Draw all icons of this type
again: afterwards its Took changes by the icon's difference and its
stats show the new icon time.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The UGC page's assemblies (cars and rockets) list had its own columns.
It now has the models' columns and sorts, minus Saved: ID and Owner (the
newest build and its creator, and how many other owners), State, Made,
Took, CPU and RAM (the latest make of any build, and the cost of the make
the UGC server did for the combination; builds that shared the made icon
cost nothing), Size (the module count), File (the build type and its
modules) and Where (how many builds and owners use it, opening
References). The gallery sorts the same way.
Check: UGC page, Cars and rockets, List: every column sorts both ways on
the server, and the gallery's sort list has the same sorts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Cars and rockets built before builds were recorded were saved with
subkey 0 and have no ugc_modular_build row, so the client has no
blueprint id to ask for their icon with. The runtime fix only reaches
characters that load again.
Migrations mysql 98 and sqlite 81 (98/81 because the guilds work takes
97/80) run ModularBuildIdMigration after the SQL: every saved item with
modules (x@ma) and no subkey gets what a new build gets, as the
character-load fix does: a persistent id from object_id_tracker (with
the character bit) as its subkey and a ugc_modular_build row with its
modules and the character as owner. The XML is written with
UpdateCharacterXml. Tried on a copy of a server's database with a
couple of thousand such items: every one got an id and a build row, in
seconds.
Check: after the migration, old cars and rockets show their icons in the
backpack (the UGC server makes them like any build) and still work
(equip, race, launch).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Live wrote the item component's UGC info on every construction of an
object with an item component (99 of 99 constructions of LOT 14535, a
possessable with item, skill and controllable physics and no Objects row:
ug_id 0, moderation status NoStatus, no description). DLU wrote a 0 bit.
The item component is read after the skill and base-combat-AI components
(ComponentOrderVector::Initialize 0x0101f8e0; live constructions of LOT 14535
have Skill then Item). DLU wrote it before the inventory, which did not show while
it was a single 0 bit next to the skill's 0 bit, but would misalign every
component after it now that it writes more. It now comes after
BASE_COMBAT_AI. The ReplicaComponentOrderTests oracle moves the item the same
way (intended change).
Test: ReplicaConstructionTest.ItemConstructionWritesEmptyUgcInfo,
ReplicaComponentOrderTest.EveryListedComponent (oracle updated).
Check in game: with a second player watching, use a possessable / mount /
vehicle item (e.g. a racing car in the race lobby); it appears, moves and
is left normally for both. Rocket launches and property models look as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The client reads an object's factions only from the DestructibleComponent
factionList (LWODestroyableComponent::LoadDataFromTemplate 0x00c9f900: one
atol per comma-separated token, -1 kept; the faction column is not read),
and live replicated that list: [-1] on 12,829 constructions (vendors,
quickbuilds, bouncers). DLU added the faction column and dropped -1,
so those objects were sent with no factions, and the four rows with faction 6
but factionList "-1" were sent as [6].
-1 has no Factions row, so it adds no friends or enemies.
Test: ReplicaConstructionTest.TemplateFactionMinusOneIsReplicated.
Check in game: vendors, quickbuilds, bouncers and NPCs still can't be
attacked; enemies still fight players and pets; smashables still smash.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Live's simple-physics constructions (40,786 in the captures):
- Motion type Fixed (5) for objects without a motionType (38,970). DLU read
the level's motionType and, when there was none, sent 0, a value the
client's motion-type enum does not have.
- Keyframed (4) for every moving platform (1,059 of 1,059) and for models
without behaviors; Dynamic (1) for models with behaviors (655 of 666 model
constructions follow this; DLU only switched it when a behavior was added
or removed).
- No velocity for fixed objects (all 38,970); keyframed and dynamic objects
have one. DLU wrote a velocity in every construction.
A level-set motionType still wins.
Test: ReplicaConstructionTest.SimplePhysicsConstructionLikeLive (values of a
captured smashable, LOT 12266); the existing SimplePhysicsTest bytes are
unchanged.
Check in game: moving platforms (AG, GF, NS, NT), spinners and other
simple-physics objects move and collide as before; smashables and static
objects stay put; on a property, placed models with and without behaviors
sit and run as before, including after picking them up and placing again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Live wrote the controllable-physics cheat block (gravity scale, speed
multiplier) in a construction only when one of them differed from 1: 216 of
20,617 constructions had it (players at run speed 1.05 / 1.5 / 2.0 / 0.5,
enemies with gravity 0), never 1 and 1. DLU wrote it in every construction.
Serializations are unchanged.
Test: ReplicaConstructionTest.ControllablePhysicsConstructionWritesCheatsOnlyWhenChanged.
Check in game: with a second player watching, use a speed boost (or /setspeed)
and zone in while boosted; the other player sees you move at the right
speed. Enemies with gravity 0 (e.g. Sentinel Turret 6254) behave as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Equipped items in the inventory component construction, as live wrote
them (49,534 equipped items in the captures):
- is_bound is the item's bound state (false for 2,990 items that bind
neither way); DLU always wrote true. NPC items are bound when the item
binds on pickup or on equip (2,640 of 2,640).
- inventory_type is the item's inventory, left out for ITEMS: TempEquip for
temporary equips and proxies (3,899), Model for models (232). DLU never
wrote it. DLU keeps player proxies in ITEM_SETS; they are sent as
TEMP_ITEMS, like live.
- An NPC item's slot is its slot in the inventory it would be in: models
count from 0 in MODELS and proxies from 0 in TEMP_ITEMS, so a proxy no
longer takes the next ITEMS slot (1,475 of 1,491 NPC constructions match;
DLU's numbering matched 1,352).
- equipped_model_transforms is an empty list on every construction (11,214
of 11,214); DLU wrote none.
Test: InventoryConstructionTest.NpcItemsLikeLive (NPCs 7426 and 13790 from
the captures), InventoryConstructionTest.PlayerItemsLikeLive.
Check in game: NPCs with gear (Numb Chuck in FV, the Ninjago ninjas, faction
vendors) show all their gear; your own and other players' equipped items,
proxies (ninja hoods, capes) and temporary equips (Maelstrom vacuum, quest
items) show normally after zoning and after equipping/unequipping.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Live wrote all three optional blocks of the character component in every
player construction (9,726 LOT 1 constructions: gm_pvp_info with is_gm
false and gm_level 0 for civilians, current_activity Some(None) on 9,620,
social_info always Some with guild 0 and an empty name). DLU wrote each one
only after something set its dirty flag, so civilians got none of them.
Serializations are unchanged (still only when dirty).
Writing the social block every time exposed two bugs in it: the guild ID
was never initialized (now 0), and the guild name was written with
sizeof(wchar_t) (4 on Linux) bits per character, reading past the end of
the UTF-16 string; it is now written as 16-bit characters.
Test: ReplicaConstructionTest.CharacterConstructionAlwaysWritesGmActivityAndSocialBlocks.
Check in game: log in as a civilian and as a GM, with a second player
watching; both see each other normally (name tags, GM tag/level for the
GM, no PvP flag), and starting a quickbuild or pet taming still shows the
activity to the other player.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Live sent the milliseconds since the object was created on the server in
every construction header (nonzero in 142,258 of 145,677 captured
constructions; the same object constructed again later carries a larger
value by the time elapsed; the local player's own construction on load is
0). DLU always wrote 0. The client stores it in the object's config
(LWOGameObject::time_since_created_on_server).
Entities now remember when they were created (steady clock) and write
their age.
Test: ReplicaConstructionTest.TimeSinceCreatedOnServerIsTheObjectsAge.
Check in game: zone in and walk around several worlds; objects, NPCs,
enemies and other players appear and animate as before (no objects
popping, stuck or misplaced animations on platforms/spinners).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
