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>
The UGC page's Make again buttons (a model's, the failed ones, everything) use
the ray tracer, hidden-face method and denoising picked next to them (the
settings' when left alone). The List view's new Options column shows what made
each model and what its next make will use; "Processing options compared"
shows every combination the made models used with its makes, models and the
average time, CPU time, hidden faces', occlusion's and icon's time and share of
triangles removed per make.
Check: the three selects next to Make again (models only); make a model again
with embree fast and see its Options column and the comparison table.
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>
The dashboard snapshot counts guilds whose name waits for moderation; the moderation counts (WebSocket and
/api/moderation/counts) include it, the Review Queue badge adds it for staff with guilds_manage, and the Guilds menu
entry has its own badge. Check: create a guild with a name off the allow list: both badges go up; approve it: they
go down.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each WorldServer_<zone>_<clone>_<instance>_<time>.log (and older WorldServer_<zone>_<instance>_<time>.log) file set
made its own server entry; they're now grouped under WorldServer_<zone>, and the file picker still shows the full
names. Check: the Server picker lists one entry per zone and its file list holds every clone and instance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A Guilds page (Moderation, permission guilds_manage, GM 5 by default): every guild with its name status, member count,
leader and age, a pending-only switch, and per guild its members (with ranks) and history (guild_events). Staff can
approve a guild name that waits for review, reject it (the guild becomes "Guild <id>"), rename a guild (same name rules
as the game, unique without regard to case), remove a member (a leader's guild goes to the next member, the last
member's guild is deleted) and disband a guild. Every change is audited and added to the guild's history, and the chat
server is asked (GUILD_CHANGED through master) to tell the online members. Pending guild names also show in the Review
Queue. Guild chat ("guild" in the chat log) counts as private chat like whispers and team chat (chat_private), with its
own channel filter and Prometheus counter.
Check on the dashboard: /guilds lists a guild made in game; approve/reject/rename/remove/disband update the in-game guild
window and name billboard of online members; the Review Queue shows a guild whose name waits for review.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Connects GuildManager to the chat server (ChatGuilds): the client's GUILD_INVITE, GUILD_INVITE_RESPONSE, GUILD_LEAVE
and GUILD_GET_ALL (routed through its world), the worlds' GUILD_CREATE, GUILD_KICK, GUILD_SET_RANK and GUILD_DISBAND,
guild chat (GENERAL_CHAT_MESSAGE in channel 10, sent to the online members as channel 10 private chat, which the
client's guild tab shows, and logged as "guild"), guildmates told when a member logs in, out or changes worlds, and a
member's pending invite dropped when they log off. Client packets go through the member's world as WorldRoutePacket;
GUILD_GET_STATUS goes to the world itself.
Guild names: the chat filter's deny list refuses a name (dChatFilter::HasDenyList, since without a deny list the deny
check refuses everything); a name the allow list doesn't cover waits for moderation, as pet names do.
The dashboard's new player action GUILD_CHANGED goes from master to the chat server only (answered 1 when chat is
connected), which catches online members up. chatconfig.ini: guild_max_members (100) and guild_invite_timeout (600
seconds).
Check in game: nothing yet on its own (the world side makes guilds reachable). Server log: the chat server starts and
logs no unhandled guild packets.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The UGC Search page and the UGC page searched the same things (owner,
account, property, the name and description a player gave a model, LOT,
UGC / blueprint id) with the same SQL. The UGC page's lists are now the
one search:
- models' List view has a Where column: placed on a property (with the
name given there and a link to the model in the property's 3D view),
in a mail or in the creator's inventories, or not found
(/api/ugc?where=1, UgcLinks::Whereabouts as before)
- owners link to their character and, with accounts_view, account
- while searching, the kind buttons say how many models and how many
cars and rockets match (/api/ugc "matches")
- /ugc_search?q= redirects to /ugc?view=list&q=; the menu's UGC Search
entry is now UGC, which opens the UGC page (it was only under Server
Admin, for ugc_manage)
- References lose their Find button (Where it is has the same) and name
the inventory a build is in
The ugc-search page and script are removed; /api/ugc_links/search stays
for API users.
Check: the menu's UGC entry; an old /ugc_search?q= link opens the UGC
list with the search filled in; owner:, account:, property:, name:,
lot: and id: searches, and the counts on the kind buttons; the Where
column's links (property, 3D, character, account).
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>
The card had the UGC page's counts per state for both kinds, totals since
the server started and the last makes, which the UGC page's lists cover.
It now says whether the UGC server is up (throttled, paused), how many
items wait and failed (linking to the failed ones), its busy workers,
CPU, memory and stored files.
/api/diagnostics/ugc read the database on a worker thread; the counts are
now read on the web thread and the worker only fetches the UGC server's
status.
Check: Diagnostics with the UGC server enabled: the card is short, reads
right while it's up, throttled and stopped, and the failed link opens the
failed ones.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
LWOSCENEID and LWOSCENEID_INVALID move from dCommonVars.h to
dZoneManager/LWOSCENEID.h; only the zone code uses them. The layer half
is the scene's layer, now the new eSceneType (dCommon/dEnums) instead of
a uint32_t: General (0), Audio (1, the *_audio.lvl scenes named
"Audio") and FX (2, Nexus Tower's *_fxs.lvl scenes named "FXs"), the
three layers the zone files of every client on disk use. ZoneScene's
sceneType is an eSceneType too, so a scene's LWOSCENEID is built from it
directly, and the dashboard compares against eSceneType::General instead
of 0. The zone checksum still hashes the layer's number.
Check in game: every world loads (the zone checksum the client checks is
unchanged), Nexus Tower included.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ghosting_scenes defaults to 1 in the worldconfig.ini template, the settings
catalog and when the key is missing. Zones without a terrain scene map keep
distance ghosting; ghosting_scenes=0 turns it off everywhere.
Check in game: with a fresh worldconfig.ini the world server log says
"Scene ghosting is on" in Avant Gardens; objects appear by scene (the whole
scene you are in and its neighbours) instead of within 100-150 units. In a
zone without a scene map (e.g. a property) the log says it is off and
ghosting is by distance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With the menu hidden on a wide screen the page had no name or link
home left; the top bar now shows "DarkflameServer" (linking home) then,
as it already does on narrow screens.
Check: hide the menu with the top bar's menu button on a wide window:
"DarkflameServer" appears beside it and opens Home; showing the menu
again hides it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The account page throttled its account_notes watcher around the check
of which account changed, but a throttled call gets no event, so every
change threw "Cannot read properties of undefined (reading 'id')" and
the history never reloaded. The check now runs first and only the
reload is throttled.
Check: lock or unlock an account (or add a note from another tab): the
history list on its page updates, and the browser console shows no
error.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Actions on an account (ban, mute, lock, GM level, email) and on a
character (edit, restoring a version, saving its XML) show the result
in place (reloadInPlace) instead of reloading the page. Deleting an
account goes back to Accounts without leaving the deleted account in
the history. After an in-place update (dash:refreshed) the pages show
the server's values again the way they do on load: last logins and the
mute end in the browser's time, and the GM level picker (unless a new
level is being picked).
Check: lock and unlock an account: the badge changes with no reload.
Change an account's GM level from another tab: the first tab's level
picker follows. Edit a character's items: the page updates when the
dialog closes. A character's last login updates when they log in.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
reloadInPlace() and goTo(url) (common.js) show the page again or open
another one through nav.js, or reload/load normally without it.
Resolving a bug report updates the page in place; deleting a bug
report or play key goes back to its list without leaving the deleted
page in the history. The System Log's server and file pickers and its
Refresh button swap the page in (Refresh keeps the scroll position).
The AI helper's "open the account" goes through nav.js too.
Check: resolve a bug report: the Resolved badge and text appear with no
reload. Delete a play key: back on Play Keys, and Back doesn't return
to the deleted key. System Log: picking a server or file and Refresh
change the log without a white flash.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Nav.refresh() fetches the current page and compares the server's HTML
now with the HTML the page started from; only what differs is patched
into the live page (text, attributes, class changes, rows added or
taken away). Typed input stays (the browser keeps what the user
changed while the default value follows the server), and parts built
by the page's scripts are left alone. Where a patch would lose
script-built or script-bound parts, or the page's layout changed, the
page is swapped in again with its scripts, keeping the scroll
position, open tabs, open folding parts and typed input; while the
user is typing it offers a refresh instead.
Live.refreshPage (account, character, bug report and play key pages)
uses it instead of reloading the page, and the "This changed while you
were editing" banner's Refresh does too.
Check: open a play key; in another tab change its uses or active
switch: the first tab's values and badge change without a reload,
and notes typed there stay. Same on a bug report page with a half
typed resolution when it's resolved elsewhere. On a character page,
an in-game save updates it without losing the open tab or scroll.
Test: NavRulesJs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
nav.js (loaded on every signed-in page) follows links and GET forms to
other dashboard pages by fetching the page and swapping in its <main>,
title, page styles and page scripts; the sidebar, top bar and live
WebSocket stay. History entries, back/forward (with scroll positions,
also kept for a reload), deep links, #hash filters and breadcrumbs keep
working. What the old page set up is taken down first: its
document/window listeners, setInterval timers, Live watchers and
topics, DataTables, dialogs and what it appended to <body>. Page
scripts run again in order; DOMContentLoaded/load handlers they add run
once they have all run. A thin bar shows while loading; a failed fetch
shows an inline error with Try again / Open it normally.
Falls back to a normal load for anything that isn't a signed-in
dashboard page, when the account or permissions changed, and for pages
with module scripts or an import map (World 3D, property view and 3D,
UGC server) or data-nav="reload", and when leaving those. Unsaved-change
prompts (beforeunload) are asked before swapping.
Check: click around the sidebar and into accounts/characters: no white
flash, the footer's live indicator stays "live", back/forward return to
the same scroll position, breadcrumbs follow the path (Accounts > an
account > a character). Activity Log links with #search= still filter.
System Log's server picker works. Leaving Settings with a change asks
first. World 3D and a property's 3D view still load (normally). Pages
that poll (Server Health, Instance Load) stop polling once left
(browser network tab). Test: NavRulesJs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The menu's groups stay open or closed from page to page
(localStorage "dash.sidebar", this browser only), applied by an inline
script right after the menu so the first paint already shows them. The
group of the current page opens and stays open until it is closed. A
new top bar button hides the menu on wide screens; that choice is
applied in <head>, also before the first paint.
Check: open and close a few menu groups, switch pages and reload; the
groups stay as left, with no flicker. On a wide window the menu button
left of the user menu hides the menu, and it stays hidden across pages
and reloads. Narrow screens: the Menu button still slides the menu out.
Test: SidebarStateJs (ctest, needs node).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Adding or removing a grant sends the existing PLAYER_ACTION refresh through master to every world: REFRESH_ACCOUNT
for an account's grants, REFRESH_CHARACTER for a character's. A world with that account's sessions (or the loaded
character) drops the grants it kept, so the next command reads them again: no relog. The dashboard toasts whether the
player was online. The dashboard's own open pages already get the new rights on their next request, and their
WebSockets are checked again.
Check: with a character in game, grant it /spawn on the dashboard (toast: "Applied in game at once") and use /spawn
without relogging; remove the grant and /spawn is refused again at once. With the player offline the toast says it
applies when they next play.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A Grants tab on the Permissions page (with grants_manage: every grant in force, the history, and a form that asks
whose it is, searching accounts or characters by name), and a Permission grants card on account and character pages
(the account's or character's grants and history; the form with grants_manage). The form picks the kind (dashboard
permission, in-game command, every permission of a category, every command up to a GM level) and searches what to
grant among only what the signed-in user may grant; grant or deny, an optional expiry and a note. In-force grants have
a Remove button when the user may remove them. Players see their own grants, read-only. The Permissions page (and its
menu entry) now opens with permissions_manage or grants_manage; the GM level tabs still need permissions_manage.
Check: as GM 9, add a grant and a deny from the Permissions page and from an account and a character page, with and
without an expiry; remove one; the lists and history update (also in a second tab). As a GM 8 given grants_manage:
only the Grants tab shows, and only permissions and commands GM 8 has are offered. As a player: your own account page
lists your grants without a form.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
What someone may do on the dashboard is now their GM level's permissions plus the grants on their account, minus its
denies (PermissionGrants.h). A deny beats a grant; denies never apply to GM 9, and settings and permissions_manage stay
GM 9 only. The account's grants are read with every request (like its GM level), so a change applies at once, and
they are passed through every check: RouteUtils::Can, CanViewCharacter, the rank rules (self_* and manage_equal_rank),
routes guarded by a permission, the templates' `can`, the API documentation, API access, API key scopes (a key never
does more than its owner may now) and WebSocket subscriptions.
New permission grants_manage (GM 9 by default) and the API to manage grants: GET /api/grants/catalog, GET /api/grants,
POST /api/grants, POST /api/grants/:id/remove. Nobody grants or takes away what they don't hold themselves (a
permission, every permission of a group, a command they may use, every command up to their own GM level), and only on
accounts the rank rules let them manage (their own with self_moderation). Commands with a fixed level or a floor
above GM 1 (/execute) can't be granted. Every change goes in the audit log (grant_permission, deny_permission,
remove_grant). Also: the Showcase gate and the traffic subscription now check their permission by name.
Check: grant a GM 2 account accounts_ban (it can ban, and the Ban button shows); deny a GM 8 account accounts_view (the
accounts list is refused); give an expiry a minute ahead and see it stop; try to grant a permission your account
doesn't have (refused); dWebTests PermissionGrantsTests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Three sections with one-line explanations instead of paragraphs:
- Test a message: normal or best friends' free chat, the verdict, each
word with why, and Block/Allow/Remove per word.
- Staff lists: one table for Blocked and Allowed with search, a list
filter and 50 per page. Block, Allow, move and Remove open a
confirmation showing where the word stands and the recent chat the
change affects (the old "What would blocking it stop?" checks).
- Word files: each file's size in one line; chatplus_en_us.txt words
searchable, 200 per page with Previous/Next; click a word to test it.
All existing APIs are unchanged and still used. docs/Dashboard.md
rewritten for the page.
Check: /chat_filter: test a message in both chats, add, move and remove
a word (confirm and cancel), search and page both lists, copy file
words into Allowed, Re-apply in running worlds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GET /api/chat_filter/test?message=&chat=normal|free says whether the
filter would stop a message from a player below GM 2, word by word and
why: blocked or allowed on the staff lists, in chatplus_en_us.txt, an
approved character name, not allowed, in blocklist.dcf, or no
blocklist.dcf (free chat then stops everything). It follows
dChatFilter::IsSentenceOkay: words split at spaces, normalized the same
way. ModerationTools::ExplainMessage, unit tested.
Check: dWebTests ChatFilterWordsTest.Explain*; the API with a word from
each list, in normal and free chat.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New Mail page (/mail, characters_mail, under Moderation): every in-game
mail newest first and live, with sender and receiver linked to their
character and account, the attachment with its icon and name (waiting
or claimed), and the state (unread, read, deleted by the player with
the time). Filters: state, character (sent or received), account,
text. Open shows the body, the attachment's item ID, subkey and data.
Locale keys in mail (%[MissionEmail_12_subjectText], the game's sender
name) are shown as the client shows them, from locale.xml; the stored
text stays under "Stored text". LocaleText::Expand, unit tested.
The character Mailbox uses the same view. Staff see mail the player
deleted (marked) and a link to the character's mail on the Mail page;
the owner sees only what is still in the mailbox, without account IDs.
/api/characters/:id/mail keeps its old fields and adds the new ones.
Check: /mail with each filter, Open on game, staff and player mail,
a mission mail's subject translated; a character's Mailbox as staff
and as the owner after deleting a mail in game.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The UGC page's model view now draws the glitter sparkle shapes (look
SPARKLE) flashing as the client's Distortion Directional shader makes
them: two layers of sparkles sliding at the client's rates (a tile in
24 s at three quarters the scale, a tile in 48 s), shown only where both
have one. Flecks are drawn as flat flakes of the fleck size and varied
brightness (addGlitter in scenery-core.js), no longer drifting (the game
never moves them). The LXFML views (the UGC page's second view, the
property and zone views) draw both on the glitter colors from
window.LDD_GLITTER, which now carries every glitter setting and is no
longer cached, so changing a glitter setting on the settings page
previews on any model's LXFML view without making it again.
Check in the dashboard (not the game): open a glitter model on the UGC
page; both views show still flecks and sparkles flashing on the glitter
bricks only; change glitter_sparkle_amount / glitter_fleck_size in the
settings and reopen the model: the LXFML view follows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The fleck texture was 50 soft round blobs of one size, bright in the
middle and fading out, which read as a smudgy dot pattern. LEGO's glitter
bricks have many small flat flakes (about half a millimetre) of which
most look faint and a few catch the light. The flecks are now flat
flakes of glitter_fleck_size (0.05 model units, i.e. 0.5 mm; 0.7 to 1.3
of it) with a one-pixel edge, each as bright as its facet catches the
light (0.3 to 1 of glitter_fleck_opacity, 80%, weighted towards dim),
80 a tile by default. The texture grows (128 to 512) to keep a fleck 3
pixels wide. New settings glitter_fleck_size and glitter_fleck_opacity;
glitter_density defaults to 80. Only glitter output changes.
Check in game: reprocess a glitter model; close up, the flecks are small
crisp flakes of varied brightness, not blurry dots; from a few metres
the brick still reads as its color with a fine glitter.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Why the glitter never moved: player models (LOT 14) are wrapped in
weeblewobble.kfm (RenderComponentWrapper 9845), so the client makes them
an LWOSkinnedRenderComponent, whose Run (0x00d6d3d0) updates the scene
graph (and so any NiTextureTransformController) only while animation is
enabled, and LWOModelBehaviorComponent::EnableAnimation (0x00be2740)
turns it off for modelType 2, which every placed property model is. The
root flags 0x102 added earlier are only read by the base render
component. Nothing in a placed model's .nif can move.
What does move: shader classes set globals in their own per-frame Run.
Distortion Directional (Ocean) (mapShaders 79, Run 0x010b90c0) slides
its texture layers by fixed shares of a tile a second, as the game's own
pond ripples (S79__pond_ripplesShape). Glitter bricks now get a sparkle
group, S79_GlitterSparkle_Model: their triangles lifted 0.005 off the
brick, vertex colors white tinted by the brick, UVs placed per brick,
alpha tested (ShaderCommon's alpha test phase, GREATEREQUAL 127), with a
stored texture of flat sparkles at alpha 230: one layer's sparkle alone
averages under the test, two meeting pass, so sparkles flash and go out
as the layers cross. The flecks stay (LEGO-AnimUV, now without the
controllers and flags that never ran). The icon and the dashboard's 3D
view leave the sparkles out.
New settings: shader_glitter_sparkle (79, 0 off), glitter_sparkle_size,
glitter_sparkle_amount, glitter_sparkle_tint, glitter_sparkle_brightness;
glitter_speed is now how fast sparkles flash (the sparkle tile). Only
glitter output changes; non-glitter models are byte-identical.
Check in game: reprocess a property with glitter models, then look at
them from a few angles and distances, on each graphics quality:
- sparkles flash on and off all over the glitter bricks, continuously
- no flickering fight between the sparkles and the brick surface
- transparent glitter bricks still see-through, flecks still visible
- nothing drawn where there is no glitter brick; icons unchanged
apart from the flecks
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every glitter brick had the same flecks in the same places: the UVs were
the vertex positions projected on an axis plane, so bricks a whole tile
apart (and every brick of the same shape at the same spot in its own
model) looked identical. Each brick now has a number of its own
(UgcGlitter::BrickSeed, from the model's id and the brick's index, kept
per vertex in Mesh::brickSeeds) that turns the projection by an angle
and moves it by an offset under a tile, differently for each axis
plane. A model made again gets the same patterns; every LOD of a brick
the same one. The icon now draws the flecks on the UVs the .nif has
(Mesh::uvs read back by FromNif). New setting glitter_random (1; 0 puts
the same pattern on every brick as before). Non-glitter models are
byte-identical (hash tests unchanged).
Check in game: reprocess a model with several glitter bricks of the
same shape; the fleck patterns differ from brick to brick.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Drops the analysis views (objects, loot odds, missions, skills, behavior
trees, activities, zones), the name search and the 3D preview, with
their API routes and CDClientRules.h. What stays: the table list,
paging, sort, any-column search, column filters, and linked values that
open the target table filtered to that ID. Old links such as
/cdclient#/object/<LOT> (UGC page, world view) now open the Objects
table filtered to that LOT.
Check in the dashboard: CDClient Browser lists every table; open one,
sort by a column, add a filter, search, page; click a LOT or loot
matrix value; /cdclient#/object/1727 opens Objects id = 1727.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The three pages had in-page tabs linking to each other, the same links
as the Logs & Health sidebar group. Removed the tabs (health_tabs
template); each page is still in the sidebar. Diagnostics gets an
<h2> heading like the other two, since the tab was its only title. The
log pages had no tabs.
Check: Server Health, Instance Load and Diagnostics open from the
sidebar with no tab row; Diagnostics shows its heading; the range and
server controls still sit on the right.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The page held one-off repairs from Nexus Dashboard (approve known pet
names, find/delete orphaned pet names, remove every buff, fix property
clone IDs, list mission rewards without a commendation price). Removed
the page, its sidebar entry, its /api/maintenance/* routes, the
`maintenance` permission and the two database calls only it used
(GetAllPetNames, FixPropertyCloneIds). The scheduled tasks (lift expired
bans, fill in pet owners, approve known pet names) and property model
import/removal stay.
Check: the Admin sidebar group has no Maintenance link; /maintenance is
a 404; the Tasks page still lists "Approve known pet names" and "Fill in
pet owners"; property import still works; the Permissions page no
longer lists "Data maintenance".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- IUgc::GetUgcProcessTotals sums what the UGC server made, for models and
for car and rocket builds: count, time spent (and how many are timed),
CPU time, average and slowest, average and most memory (estimate),
bricks, triangles and triangles saved (models whose count before hidden
face removal is known). /api/ugc returns them as totals and the UGC page
shows a card for each kind. Parity tested.
- Player models no longer have a shared icon preset: every one is a
different size and shape, so its icon is fitted to it from the settings.
The UGC server ignores a kind:model preset, the dashboard refuses to save
one, and the icon editor hides the type buttons for models. One model's
own icon values still work. Car and rocket build types keep theirs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Purge all ran on the UGC server's main thread inside the dashboard's
request: it measured and deleted every stored folder, then updated rows one
by one, and the main thread (downloads, the master link, the database)
waited minutes for it. Now a purge answers at once and a file thread lists,
measures and deletes the folders; the main thread applies what it removed
and updates the rows a few hundred a tick. No new item is started while a
purge runs, and items being made when it starts are left alone, so no worker
writes into a folder being deleted. One purge at a time. The storage cap's
clean-up lists and deletes on the same thread. The UGC page shows the
purge's progress (/status purge).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
fmt.property(id, name) links a property's page and adds a small 3D button
straight to its 3D view. Used everywhere a property is linked: the
properties list and top slots, moderation, character pages' related
properties, UGC search and the UGC page's placements, the home page, the
chat log, player reports and report notes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Flair tints are the terrain file's color bytes over 255, times the flair
model's own vertex colors (FlairAssets::AppendRenderBuffer divides by
255.0). The manifests said 63, so in the game shaders' sRGB maths tints
of up to 4 turned the flairs white (Battle Against Frakjaw's leaves).
Conversion format 6, so flair manifests kept by browsers are fetched
again.
- BasicShaders, AlphaAsAlpha and Flair.fx multiply the light by the vertex
color and only then clamp, as the COLOR0 output; the views clamped the
light first, so a light brighter than 1 no longer lifted darker vertex
colors.
- Docs: the flair tint, the clamp, and that AlphaAsAlpha cards show their
backs (Cullmode none) from outside a playable area.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The game shader views looked up each shader's technique in the manifest's
"techniques". Property scenery manifests are cached by browsers for a day
(world ones for an hour), so after the update a browser drew the property
view from the manifest the older server had sent, which has no
techniques: every shader fell back to LEGO, whose decal texture alpha
laid the see-through tree, rock and water textures over white vertex
colors. Nimbus Isle came out with white trees, rocks and water, a yellow
build surface and a solid white build border.
- Manifest URLs carry the conversion format the views are written for
(?format=5, scenery-core.js SCENERY_FORMAT), so a kept manifest from
an older server is never used; SceneryCoreJs checks it matches
Scenery.cpp FORMAT_VERSION.
- A manifest without techniques (an older server's) is drawn with the
viewer's own lights and its textureAlpha table instead of every
shader guessed as LEGO.
- A material whose NiAlphaController animates its alpha is drawn at its
highest key. The AnimAlpha shaders now use the material alpha, and
effects resting at 0 in the file (the Venture Explorer's lightning)
had vanished. Conversion format 5.
Checked by rendering the world view of every zone with models and the
property view of every property template (headless, fixed cameras)
before and after, and the Nimbus Isle property with a manifest stripped
of its techniques, which reproduced the white look.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
game-shaders.js ports the client's techniques to one ShaderMaterial
program per family and variant: LEGOPPLighting (hemisphere lit sun,
ambient, fresnel rim, N.H^320 specular, the default reflection cube;
decal and non-decal textures, emissive, super emissive, glow, grayscale,
no ambient, AnimUV), BasicShaders and AlphaAsAlpha (lit or unlit, both
sides, alpha blend, alpha test, additive, animated alpha, two layers),
Metallic.fx polished metal and brushed steel with the client's metal
cubes, clear plastic, the Distortion (Ocean) layers, Flat Surf,
BrickWater, darklings, terrain meshes, flairs and the sky. The maths runs
on sRGB values as Direct3D 9 did; lights, specular, hemisphere and fog
blend between scenes; textures move by the .nif's texture transforms.
Programs are shared by every material with the same defines.
A Fog switch (off by default) adds the zone's fog to scenery and
terrain. Shaders the game doesn't draw in the world (footprints, drop
shadows, post-processing) are left out. Docs list the families, their
gameValues and what's approximated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NifFile::TechniqueFor maps every mapShaders gameValue to the client's
technique family (fixed function, LEGO, Basic/AlphaAsAlpha, metal, clear
plastic, ocean distortion, flat surf, BrickWater, darkling, terrain mesh),
its eShaderLook bits, texture alpha and eTechniqueFlag bits (moving
texture, both sides, blend, alpha test, additive, no ambient, glow,
super emissive, grayscale, shiny glint, not drawn, ...), from res/shaders
and the verified technique setups. Values it lacks are the LEGO shader,
as the client falls back to it. TextureAlphaFor and ShaderLookFor read it.
The scenery manifests carry it as "techniques" (replacing textureAlpha
and shaderLooks), the flairs' manifest a Flair.fx technique, the
lighting its specular color. /api/scenery/env/:name serves the
environment cubes the client's shaders load themselves (default
reflection, polished and brushed metal, brushed noise). Conversion
format 4.
scenery-core.js: techniqueOf, gameLook with the family and flags,
blendingOf, parseDdsCube; its test checks the flag and look bits against
NifFile.h.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The 42 depth renders only saw faces in direct view, so faces reached only by
bounced light (interiors, recesses, rooms seen through openings) were removed.
UgcHsr traces the paths LU Toolbox's Cycles bake traces, directly from points
on each opaque triangle, and removes a triangle only when none of its paths
reaches the sky:
- points in rows along the triangle's longest side, hsr_sample_spacing apart
(0.1143, 7 x 7 on a stud-sized square), at least hsr_min_points (28, the
texels LU Toolbox bakes for a triangle), at most 4096
- hsr_samples (8) paths from each point, at most hsr_bounces (8) bounces, as
Cycles 3.1 samples the bake material (Principled BSDF defaults: Burley
diffuse and GGX specular, defensive sampling, Filter Glossy, 4 glossy
bounces, Russian roulette from the second bounce, ensure_valid_reflection);
no direct sky sampling (Cycles doesn't sample a flat world as a light)
- hsr_ground_plane: LU Toolbox's black box under LDD's floor
- decided per triangle, deterministic per model; triangles without area removed
- the VC pre-pass isn't done
Checked against LU Toolbox's operator in Blender 3.1.2 on the same meshes (27
models, 937,814 triangles): 501,362 removed there, 501,172 here, differences
as large as LU Toolbox's own between seeds. optimize_resolution is retired;
remove_hidden_faces=0 makes the same files as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With new binaries in place, master moves everything onto them while the
server keeps running (the dashboard's Live update, /liveupdate or SIGUSR2 to
master):
- database migrations of the new build first; a failure stops there
- UGC finishes the jobs it is running (queued rows stay pending), auth
restarts, chat hands its teams to master for the next chat server; master
starts the new processes and retries ones that don't come back
- once the new chat server is up (CHAT_SERVER_READY) every world connects at
once and sends its players again (LoginSessionNotify resync, no login logged)
- every world instance is replaced with an instance migration: public worlds
and private ones (same password) at once, properties after the old instance
saved and froze the property (MIGRATE_PREPARE: no building, claiming or
saving there any more), activity zones and character select once their
players left or after a wait; empty instances just stop, zones in
prestart_worlds get a new one first
- the dashboard restarts last and picks the status up again
Players land where they stood (position carried in CarriedPlayerState, also on
properties and Moon Base). Draining instances get no new players
(InstanceMigration::AcceptsNewPlayers) and show as "Moving players" in the
world list. The order lives in LiveUpdateMachine.h without master state and is
unit tested; master's glue is LiveUpdateCoordinator. Master itself is not
replaced. Message IDs are appended only.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
"Clear threat list Trigger Wall" (LOT 13632) on the NS medium property was
drawn as a big red wall: nothing in its data hides it, its client script
does (scripts/02_client/map/general/l_set_invisible.lua sets it invisible in
onRenderComponentReady). The scenery now reads each object's client script
once and treats an object as hidden, like renderDisabled, when a
self:SetVisible{visible = false} is directly in onStartup or
onRenderComponentReady (not under a condition). 12 LOTs match: trigger
volumes, effect containers, a scripted camera, dummies, booty chests (shown
once dug). They show with Hidden objects on.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
