The suite spent 88% of its step time asleep: 564 s of steps, 496 s of
them in sleep(n) - fixed spans estimated by hand and then copied from
step to step. That is why it was slow, and why it flickered.
The signal is afterFrames(count, timeout) in ui/render.ts, exposed as
__gm.afterFrames(). It wakes the loop itself, because an editor that
draws nothing would never redeem a promise of a frame, and it gives up
after a timeout instead of hanging the run. settle() in the harness
waits two drawn frames (~32 ms) instead of 150-300 ms and falls back to
sleep(150) when the hook is missing (an old build): silently not
waiting would make every later check a coin toss.
Every action goes through it now - click, doubleClick, wheel, drag,
dragAsset, press, typeText, select, mutate, undo, redo - and with that
about 550 sleep() calls behind those actions became redundant, helpers
that are nothing but a mutate in a coat included. What a frame cannot
answer got its own condition instead: waitUntil(expr), play()/stop()
waiting on __gm.state.mode rather than on 300 ms, saved() for the
autosave (the two hand-written polling loops sampled at 300 ms for
something that arrives after a second), and showTab() as a real tool -
47 places called evaluate("__gm.showTab(...)") raw, and 24 of them
waited for nothing at all and read the old panel.
564 s -> 243 s of step time, the waiting share 88% -> 76%. Three kinds
of wait were kept, and each one bit when it was taken away: loading
bytes (textures, fonts), game time that a repaint cannot stand in for
(time scale, shader and particle clocks, the live values a running
graph carries), and the inertia the scroll step measures on purpose.
Two things the new speed uncovered, both real: 25-animation read the
frame menu unfiltered and had frozen an incomplete list (the CI failure
on c990dbc), and 33-render-on-change measured idleness while the
autosave of the step before was still painting - it waits for
saveStatus to leave dirty/saving now, so it measures that nothing
happens any more instead of that it waited long enough.
CI found the gap the hard way: with every kind of asset in the project
the new scene sits below the panel, and the right click for "Set as
start scene" landed somewhere under the list instead of on the row.
locate() only asked whether a text was visible, and a masked row still
is - it is merely clipped away.
locateAll and overlayTexts now walk the tree together (eachText) and
carry the clip of every ancestor, measured on the widget that owns the
mask: Pixi v8 hands a Graphics set as a mask to the mask effect and
reports empty bounds for it, so asking the mask itself declares every
masked text in the editor missing. Two hooks with two notions of
"visible" was the second half of the same bug.
The panel also scrolls the selected row into view now, the way the
hierarchy has always done - an asset selected elsewhere has to be
findable in the list, test or no test. And step 21 learned that scene
is a property kind.
Load scene sets a flag on the World like Stop and Restart do; the swap
happens after the tick, and the last wish of a tick wins. Exactly one
thing crosses over: the SharedState, so global variables and global
machines keep what they know while everything local is built anew.
Loading the scene that is already running rebuilds it, which is what
"reset the level" means when the score must survive.
A run starts either from the start scene (a new project field, set from
the asset panel and marked with an arrow in the list) or from whatever
scene is open. The toggle next to Play says which question is asked and
remembers the answer in localStorage: it is a way of working, not part
of the game.
Sounds no longer end because a scene ends. A voice belongs to the
behavior that started it and stops right after On destroy, so the scene
swap is no longer a special case at all. For the enemy that plays its
explosion and destroys itself in the same breath there is now a
detached input on Play sound.
The scene selector stays usable while a run plays, because a double
click in the asset panel already did the same thing unguarded. It moves
what the editor looks at, not what the run plays.
Typing "variable" into the node palette landed on `Run behavior`, because the
filter weighed label, type and description alike and then sorted by category.
`searchRank` in `src/ui/search.ts` now ranks a match: exact label or type, then
prefix, then label, then type, and only last a hit that lives in the
description alone. Category and alphabet order within a rank. The searchable
`Menu` uses the same function, so searching means the same thing everywhere.
The `+` menu grew to fifteen entries in declaration order. It is built in two
alphabetical groups now -- the types that go into a scene, then the ones that
stay in the library -- with a separator between them, the split `isPlaceableType`
already draws. `ASSET_TYPES` itself stays untouched: it is also the
serialisation and check order. A search flattens the groups and drops the lines.
Popups were capped at fixed heights (320 menu, 260 palette, 340 pickers) while
the canvas had room to spare. `Popup.roomFor()` measures its own chrome and
takes the window height minus a tenth at each end. Placement clamps
asymmetrically: the bottom margin is enforced, the top is only a budget, so a
menu anchored near the top stays at its button instead of being pushed off it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Sounds follow the road fonts already paved: the file goes to the server
content-addressed under its sha256, the project keeps a named asset that
points at it, and nothing is ever re-encoded. The format is read from the
file's own signature, not from what the browser claims - the same .m4a
arrives as audio/mp4, audio/x-m4a or as nothing at all.
Playing is two nodes. "Play sound" takes the asset and may raise or lower
its volume; every call is its own voice, so two jumps in a row are two
jumps heard. Its loop input is a choice of three - from the asset, on,
off - because a checkbox has to spend one of its two positions on "not
set" and would lose a case: an asset that loops could then never be
played once.
Below the graph core stays browser-free: NodeContext hands out an asset
id, BehaviorSystem resolves it against the project and folds the asset's
own volume and loop into the call, and only viewport/SoundLibrary knows
about Web Audio. A run without a sink plays nothing instead of throwing.
The CI browser is muted, so the ear is no witness: the debug hook counts
started and stopped voices, and step 34 reads that counter.
Four issues that all reach for the same line: how the current state is
read and written. Splitting them into four commits would mean four
states of the tree that do not compile, so they arrive together.
#55 start state — the machine asset carries an optional `start` holding
a State node's id. Right-click a state to set or remove it, a ring in
the node's title says which one it is. A dead id (the state was deleted)
reads as "no start", so nothing has to clean up after a deletion.
#53 entered/exited — `On state entered` and `On state exited`, one exec
output per state. Every change now goes through a single `setState`
that reports it; BehaviorSystem delivers it after the exec flow that
caused it, `exited` of the old before `entered` of the new, round after
round until the queue is empty, with a cap that stops a ping-pong graph
with an endless-loop error. A local machine tells the instance that
switched, a global one tells every instance holding a matching event
node. The start state fires `entered` too: per instance for a local
machine, once per run for a global one.
#57 reading — `Current state` with `name` for showing and `id` for
comparing, `Is in state` with one boolean per state. Neither has a
`run`, so `read` learned `def` and `lookup`, and the catalog test knows
this third kind of node.
#54 seeing it — during a run the state machine tab marks every state
some instance is holding faintly, and the one held by the selected
scene object's instance strongly.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The renderer is taken off the ticker; the ticker itself keeps running, so
the clock, a live world and the panels still get their step. Drawing now
happens inside a 200 ms trailing window that every pointer, wheel and key
event, every Widget.invalidate/resize, every Stack.relayout, every
EditorState.emit and both library onChange hooks reopen. The window rather
than a single frame is what makes the ~70 direct redraw() calls that never
pass through invalidate() safe, along with repaints an event only starts:
a microtask, a texture upload, a resolved font.
Editor.movesByItself() keeps asking for every frame while Play or an editor
run is on, a time shader is on screen, or the particle or animation preview
is showing.
The FPS counter now counts drawn frames over the time the renderer was
awake. Idle seconds are not slow seconds, so it answers "how smooth while
it moves" and holds its last answer instead of falling to 0 at rest.
Measured on the built stand, 20 s per variant, CPU of all chromium
processes from /proc: kept awake 75.5 % of a core at 60 fps, render on
change 8.8 % with 0 frames drawn, awake again 83.9 % as an order probe.
No colour check can see a missing repaint: pixelAt extracts from the stage
on demand and is just as green on a screen frozen a minute ago. Step 33
watches the new __gm.renderStats() counter instead.
A state keeps its `in`/`out` exec pins in the model and the editor never draws
them. That is the whole trick: connecting, undo, deleting a state and saving all
keep working untouched, because `normalizeGraph` filters nodes and never edges,
and no edge ever checks its port key against a NodeDef. Only two places change --
`NodeView.redraw` stops drawing the row, and `portAt` refuses to hand one out, so
an invisible pin cannot be grabbed either.
Transitions are drawn by right-clicking a state, `Make transition`, then clicking
the target. The armed mode takes the same shape as a live drag, so the existing
preview draws it with no second code path, and it dies with Escape, a click on
empty canvas or a switch to another asset.
Removing the pins removed the only handle a transition had: edges have no
hit-test anywhere, and the pin right-click was the sole route to Disconnect.
The inspector's TRANSITIONS OUT / TRANSITIONS IN lists are therefore not a
convenience, they are the delete path.
`GraphEdge` gained an optional `label`, which `normalizeGraph` has to pass
through explicitly or a saved name is dropped on load. `State transition`
replaces `Next state`: it takes the machine on an input port and the name from a
picker, and resolves against the *current* state. Names are deliberately not
unique across a machine -- two states may both have an `alarm` -- so a node named
`alarm` moves whichever state has one, and an empty name follows the single
unnamed edge.
The geometry moved to a Pixi-free `transitions.ts`: a gap to the box border, and
a fan that puts a pair of opposite edges beside each other instead of on top of
each other. Keeping it as plain maths is what let a screenshot bug become a unit
test -- an earlier pin-to-pin line ran under the target node and took its
arrowhead with it, because `nodeLayer` sits above `edgeLayer`.
One limit stands and is not fixed here: `ConnectCommand` rejects a second edge
between the same two states ("Already connected"), because every transition
hangs off the same two ports. With names, `A --alarm--> B` beside
`A --seen--> B` is a sensible pair, so this is worth its own issue.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Particle age and shader time each read performance.now() off their own
library, and the world got the frame's delta straight from the ticker. Three
clocks, so a global time scale would have been three multiplications and a
pause three separate freezes.
core/time.ts holds one clock instead: it accumulates dt x scale per frame
rather than reading the wall clock, and Editor.tick hands the very step it
took to the play session, the editor run and the animation preview. Delta
time, particle age, shader time and that preview are the same quantity now,
so the scale is a single multiplication and a pause is that scale at 0 —
nothing underneath needs rebasing, because the clock itself stands still.
The Math.min(dt, 0.1) caps in PlaySession and EditorRun are gone: the cap
(MAX_FRAME) now sits on the raw frame before scaling, and capping twice would
make the world run slower than the particles at scale 2.
A pause gates the scale rather than overwriting it — effectiveScale() answers
0 while paused, and for anything that is not a positive finite number, while
gameClock.scale keeps its value. A slider later survives a pause that way;
the issue asks for none yet.
The clock goes back to zero at both ends of a run, so a run is reproducible
from second 0 and the scene view looks freshly loaded after Stop. Resetting
only on Stop would not do it: the editor clock runs on afterwards, and the
next Play would start at "seconds since the last Stop".
The trap in that: ParticleField keeps its origin as an absolute second of
this clock, so after a reset it lies in the future and every particle would
age backwards. One line in sync() starts such a field over, which covers the
scene, the game and the tab preview at once — and fixes an older bug of the
same shape, where releasing a particleTime() freeze left the preview empty
for good.
Found while measuring that: ParticlePreview stood at second 0.5 by computing
the clock reading that would mean 0.5 since the field started. Over 200000
random origins the round trip misses the last bit in 18.8% of them and shows
one particle too few in 9.4%, because floor(0.5 * 20) comes out 9. The check
had been flaky since it was written. The panel hands the field the age itself
now (ParticleField.hold), and the faked clock is gone.
833 checks (824 + 9), 931 unit tests (921 + 10).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Behavior and Container draw an icon because isSymbolType() is true for them,
but that predicate does not mean "draws an icon" — it means "is editor
decoration". Four places read it that way: collectWorldItems, GameViewport.
collectFrame and syncDrawOrder all drop symbols out of the running world and
the game picture, and placementOf() gives them an upright unscaled box.
Listing 'particles' there would have handed the source an icon and taken away
its particles, in the scene view as well as in the game.
So the two questions are two functions now. hasSceneIcon() says a type draws
an icon; isSymbolType() keeps meaning decoration. The icon itself is a
SceneViewport decoration, which is editor-only for free because GameViewport
never syncs decorations.
A ring (r 10, #c26a1f) with four sparks, upright and unscaled like the other
symbols, its middle left open — unlike Behavior and Container a source draws
something of its own, and the field is densest exactly there. The name sits
beside it and a dashed line runs to the parent, same colour.
Found while writing the tests: the source's grab box scaled with its transform
while the icon did not, so the two came apart. placementOf() returns only the
translation for 'particles', making the box the upright 24 x 24 that covers
its icon. The source's own transform still feeds every particle untouched.
824 checks (821 + 3).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three reports plus one found on the way.
Hold only froze the *picture*: the preview computed a fake second while drawing,
the library clock kept counting underneath, so letting go jumped to "how much
would have passed by now". That is the flip side of the formula model — a thing
that is a function of the clock is not stopped, its origin is moved.
ParticleField.rebase() does that; hold(null) calls it with the second stood at.
The Hold label now mirrors the preview's state instead of its own click history,
so dragging the slider (which holds too) reads `Run`, and Restart puts a held
preview back to 0.00 s rather than leaving it on the second it stopped at.
previewProps only read defaultsFor('particles'), so the panel showed the type's
defaults while the inspector showed the asset's rows. AssetBase.props is a full
bag on a graph asset as well — asset.props now lie over the defaults, the
asset's texture is resolved instead of always drawing a white square, and the
slider's span follows its lifetime. The span is re-read in tick(), because the
rows are edited in the inspector and the tab hears nothing about it.
The preview moves from under the graph to beside it, past a draggable splitter:
a graph is read across, so width is what it can spare and height is not.
AssetTabEditor grows a body() hook for that — the base class still stacks header,
strip and canvas, a subclass may put something next to the canvas.
Found by the full run, not reported: the graph canvas's empty-state hint ran
38 px past its frame now that the preview takes 280 px away. It is cut with
truncate() like any other label.
816 checks (809 + 7).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The asset type `particles` is placeable and carries a graph at the same time —
`behavior` was the blueprint. Because a continuous emitter gives birth to particle n
at time n/rate, the whole field is a function of the clock rather than of the frames:
no ring-buffer bookkeeping, exact time jumps, reproducibility as a measurable property.
compile.ts follows compileShader's frame but emits JS closures into a Float64Array
instead of GLSL, and without eval; a node with two consumers is evaluated once.
Drawing is one Pixi ParticleContainer per source, so one draw call.
Collision, wind and braking in flight are out for good — that is the price of the
formula model, not an omission.
The tab count goes from nine to ten, which is exactly the case #84 was built for one
day earlier: at 1280 px the row no longer holds every tab, so guard 14b now asks for
hidden.length > 0 instead of === 0.
Nine tabs need 631 px; at 1280x800 the centre column offers 666 px, so the tenth
tab would have spilled under the inspector - the same overflow CI caught in run 5.
TabStrip replaces the bare HStack: it fits the longest prefix that stays inside
the budget, always appends the active tab, and offers the rest behind a fixed
28 px button. The button's width is fixed on purpose - deriving it from the
number of hidden tabs would make width depend on the count and the count on the
width.
The rule itself is a pure function (tabRow.ts), so it is checked without a
browser. setActive() calls relayout() itself because Widget.resize only lays out
on a changed size, and switching tabs does not change any size.
Adding a tab now needs no space arithmetic - only a line in TAB_ORDER,
TAB_LABELS and panes, all three Record<CentreTab, ...>, so forgetting one is a
type error.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Only the picture was masked, so the checkerboard, the frame outlines, the
handles and the name plates were drawn over the panel next door once the
sheet was zoomed past the tab. The dark block in the report is the
checkerboard's base: checker.ts fills the whole image rect and clips only
the light squares to the widget.
Everything that pans and zooms now sits in one masked container, the
pattern WorldViewport and AnimationCanvas already use. A check samples
pixels outside the tab and names the frames drawn past it; a guard in
front of it asserts that zooming really pushes the sheet over every edge,
or the pixel probes would measure nothing and stay green forever.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A font asset holds a file and nothing else, the twin of a texture: the server keeps it
content-addressed under data/fonts/{sha256}.{ext}, with the format read from the file's own
four-byte signature (finfo reports fonts inconsistently across libmagic versions). A text
wears one through its ordinary Font property, so overrides, reset, undo and the Properties
node come for free; no font means the editor's own typeface.
The picker lists everything the server holds and renders each row in the face it offers.
The family name is gm-font-<hash16>, never the name inside the file, so two "Regular" cannot
collide. WorldViewport.syncText is the single place that draws it, which covers scene, game
preview and play at once.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
shaderFingerprint(graph) covers node id/type/params/ports and edge endpoints only;
x/y, collapsed, folded and edge ids stay out. ShaderLibrary keys on it and counts
compiles (__gm.shaderCompiles()). Measured before: one compile per preview step of a
node drag (12 for a harness drag); after: 0, a param change exactly +1.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Four new rows on sprite, text, collider and container (block TRANSFORM):
scale, rotation in degrees clockwise, mirrorX, mirrorY. As properties they
get override/reset/apply, loading of old projects, write-back after play and
the Properties node for free; camera and behavior have none.
The hierarchy composes T*R*S per level (composeAlongChain replaces
sumAlongChain), so children rotate and scale with their parent. Reparenting
and grouping keep the look (carriedTransform). Drawing goes through
placeNode, the gizmo is a polygon, cull and objectRect use the AABB, drags
are computed in the parent's system.
Colliders are oriented boxes: SAT for box/box, box-circle in the box frame;
exitDistance replaces axisSeparation/axisCorrection, moveAndSlide moves along
the object's own axes.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Fragment-side coordinate nodes instead of vertex nodes: the quad stays the
quad, a UV read plus a sample at a computed coordinate covers flip, scroll,
wave, pixelate and scanlines. Time is a uniform only when a Time node reaches
the Output; a viewport that drew a Time shader redraws from the ticker so the
Scene tab animates without a running world. Edge (Mirror/Repeat/Clamp) is a
second texture variant in TextureLibrary because WebGL binds no TextureStyle
resource of the shader. The compiler now emits #version 300 es first: Pixi
only compiles as ES 3.00 when the fragment says so, and textureSize needs it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A sprite's picture is now a Mesh (unit quad, sibling of its Graphics node, which keeps
only the hit area); a worn shader replaces Pixi's fragment in that draw, so there is no
intermediate texture and no extra render pass. Measured on the HD 630: 400 shaded
sprites hold 60 fps (was 34), 800 hold 56 (was 20).
Two Pixi facts the code leans on: a mesh's texture is bound only for Pixi's own shader,
so GraphShader.wear hands it over; and MeshPipe caches whether a mesh batches and
re-checks it only in a frame without structure changes, so switching a shader on or
off calls mesh.unload().
The console panel mirrors console.group/groupCollapsed/groupEnd and fills the %s %d
%o %c directives, so Pixi's deprecation warning shows its head, not only its stack.
E2E 12 guards against the addChild-on-Graphics warning, 28 counts shaders instead of
filters.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The wiki carried an estimate ("fine for dozens, not for thousands"). Measured
on the Intel HD 630 that CI uses, 48x48 sprites in a 640x360 camera, neutral
values, three runs agreeing to 1 fps:
sprites 25 50 100 200 400 800
fps 58 56 54 48 34 20
The same scene without shaders holds 60 fps at every one of those counts, 800
included. Quadrupling the sprite area (96x96) measures the same, so the cost is
the render target switch, not the pixels: roughly 20 us per pass up to 200,
40 us at 800.
tools/perf-shaders.mjs drives it (not run in CI, frame times are machine
dependent). New hook __gm.filterCount() counts visible nodes carrying a filter
and proves one pass per sprite; e2e 28 checks it.
Two traps the tool guards against, both cost a wrong answer first: the median
frame time is useless here (the distribution is bimodal under vsync, so the
rate comes from frames/duration), and --disable-gpu-vsync has no effect in
headless, so what is measured is the point where the frame rate breaks down.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A frame drawn straight from the sheet showed a hairline of the neighbouring
cell along any edge that fell inside a screen pixel: with antialias on, a
partly covered pixel is still evaluated at its centre, past the edge. On top
of that Pixi rewrites a Graphics fill texture from clamp-to-edge to repeat at
draw time (generateTextureFillMatrix.mjs), so the top row of a sheet read its
bottom row.
TextureLibrary.get(ref) now cuts the rectangle out of the loaded sheet with an
OffscreenCanvas on first use and uploads the cut (key file|x,y,w,h, kept for
the session). Address mode is mirror-repeat: Pixi only rewrites clamp-to-edge,
and within a texel of the border mirror behaves like clamp, for linear as well
as nearest. Loaded sheets get it too, since whole pictures wrapped at their own
borders the same way. viewport/textureFill.ts and its test are gone.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Dropdown labels are now Linear (default) and Nearest; the stored values
linear/nearest are unchanged, old project files load as before.
e2e 12c-ii measures in Play at camera zoom 1.37 (screen scale 1.434): a
Nearest edge sampled across seven pixels is byte-identical with and without
a neutral Adjust colors shader, and brightness 0.5 on the same path darkens,
so the pass really runs there. New hook __gm.gameRect(id).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
No input, one Value param (step 0.05, drawn on the node), output `value`
as a float literal via floatLiteral (2 -> 2.0, NaN -> 0.0). No uniform,
no sprite row - that is the difference to Property.
The e2e negative check for "no behavior node in the palette" now uses
Boolean as its marker, and step 28 types the palette search first
because the palette scrolls from twelve entries on.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The ninth tab (Shader) pushed the row past the centre column: List
ended at 962 while the column ended at 925, so the last buttons sat
under the inspector and clickText('Container') hit the List tab
instead (CI runs 5 and 6, module 16). Tab buttons now use padding 4
via a new Button padding option; e2e step 14b guards the row width
using the widget box rather than getBounds(), which sees only the
label on an untoggled ghost button.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A shader is an asset carrying a node graph (Texture color, Output, Property,
Color, Add/Subtract/Multiply/Divide, Mix, Luminance) that compiles to the GLSL
fragment of a Pixi filter. The Shader tab reuses the Behavior canvas; a sprite
wears a shader through its Shader picker, and the rows below it are the
graph's Property nodes. Property schema, resolution and deviation take the
project because a sprite's rows now live in another asset. The built-in
registry of the prototype is gone; "Adjust colors" ships as a graph in the
fixture and the three example games.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A log layer (src/editor/console/log.ts) keeps a 500-line ring buffer and
wraps console.log/info/debug/warn/error, still calling the original. Notices
and behaviour runtime errors go straight to the layer instead of through
console, so editor tests stay quiet and errors can render red. world.log is
untouched: it is per-run structured state that runtime tests and e2e read
directly, not a text stream.
The panel sits under the visible tab, resizable with Splitter('y'), with a
CONSOLE header and a clear button. Each line carries its wall-clock time in
its own dimmed column, so locate() still finds the message text.
The taller centre area drops the graph fit zoom to 0.60, which broke six e2e
checks that assumed one screen pixel is one graph unit and that a fixed
offset from the canvas centre is empty. They now ask __gm.emptySpot(),
__gm.graph.emptySpot() and __gm.graph.zoom() instead of guessing.
locate/locateAll added the unscaled Pixi text width to the global position,
so click points in the scaled graph canvas landed too far right; they use
getBounds() now.
check.sh: 831 tests. Full e2e: ALL PASSED - 671 checks. Smokes 24/19/20.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A state machine is an asset whose graph holds one `State` node per state
(no ports, name as title), edited in a new "State machine" tab that reuses
the graph canvas: right-click places "State n", snap toggle shared with the
Behavior tab via SnapToggle. `statemachine` is a value kind like texture
(port, literal node, variable, struct field, tagged list item).
Behaviors use it through `State branch` (one exec output per state) and
`Set state` (one exec input per state; the interpreter now hands `run` the
entry port). States live per behavior instance in NodeContext.states, with
no initial state. Ports derive from the inline picker or a wired
`State machine` literal; BehaviorLookup became GraphLookup with
behavior(id) and machine(id).
The Graph tab is labelled "Behavior" (key unchanged). Unit tests
(state-machine, kinds, catalog, ports), e2e step 26 and README added.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- applyCommands sit on ObjectEditCommand with a revert hook instead of
five copies of the same preamble and undo
- actionIf replaces six identical Reset/Apply menu blocks, addNameField
the three name rows, assetLink the two asset links
- requireStructuralObject replaces ten two-line preambles and drops the
duplicate holder lookup
- __gm.objectsOf had no caller; the SMB remnant in e2e.sh is gone
Proven: check.sh 797 tests, full e2e 633 checks, smokes 24/19/20 against
the same bundle.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A list no longer says what its items are. ListDef.items, ListDef.fields,
ListFieldDef.items and SetListItemsCommand are gone; the single seam is
src/core/properties/listItems.ts. JSON already tells number, text, boolean
and struct apart, so only two kinds need a tag: {"$color": n} and the new
{"$texture": id}. Field names cannot start with "$", so a tag never
collides with a struct field. On a data link a texture item arrives as
its bare id; the graph has no images.
List tab: the mark in front of each row is the kind picker, "+ Item"
asks for the kind, both are ordinary value edits (one undo step). A list
inside a struct field only offers the flat kinds because the depth rule
forbids a struct item there (ListAccess.kinds); the menu used to offer
it and silently do nothing.
Legacy lists that declared items: 'color' are tagged on load.
Actor demo: the four walk animations and four standing frames live in
two list variables of texture items indexed by facing (Item at), so the
graph can be rebuilt in the editor. The stand frames get their grid ids
back; 44 nodes instead of 46.
Also: the user's edits to on-start-during-play.md and the new note
reseed-because-of-list.md.
Verified: check.sh 794 tests, full e2e 631 checks, smokes 24/19/20.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
An animation asset is a list of pictures (whole textures or frames) with
frames per second and Loop/Once/Ping-pong playback. A sprite wears it
through its texture property, so the picker, instance overrides and the
Properties node work unchanged; there is no dedicated play node.
A different animation starts from its first frame, the same one set again
runs on, so a graph may choose Run or Idle every frame. The clock lives on
the entity, not in props, which the world writes back as overrides. The
AnimationSystem ticks before behaviors, and On animation finished fires
once when a Once animation reaches its end.
The Animation tab plays a preview above a frame strip: reorder by dragging,
Delete, duplicate, + Frames, pause/resume, and wheel zoom about the pointer
with pan and Fit. Scene and Container tabs show the first frame; edit mode
has no clock. Right-clicking a texture creates an animation from its frames.
Button now measures its label with the bold font it draws in, which cut
the Animation tab and red dialog buttons short.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The graph was only locked by five edit-mode guards in GraphCanvas. The
running game reads the graph by reference each frame, so a new node, edge,
changed value, disconnect or undo takes effect from the next frame. Stop
discards these edits like everything else changed in play. A new On start
node does not fire for a behavior that has already started.
During play every data output shows the value it last produced, in the
highlight colour and refreshed ten times a second. Values are kept per
graph, because node ids are only unique within one; with several
instances the most recent value wins. The Variables node reads the live
variable bag, so only its own outputs are kept, as a snapshot.
Each time an exec edge fires, a bead runs along it in 0.3 s, at most one
every 0.15 s per edge, following the same curve the edge is drawn with.
Beads run on game time and freeze while paused; the game logic is
unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A single node now lists any number of variables. Every entry is an output
that reads the current value whenever it is pulled; an entry switched to
Set is also an input and is written when the flow arrives. Exec pins only
exist while at least one entry writes, and all inputs are read before any
write, so one node can swap two variables.
Get/Set variable, Get/Set variables and Variable are gone. Old project
files are converted on load: each old node becomes a Variables node with
the same variables, edges and typed values.
A written port without a typed value now shows the zero of its kind on
the node body, matching what the inspector shows and what the run writes.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Get properties / Set properties work like their variable counterparts:
one port per chosen property, an unconnected input writes its inline
value. A port chosen from the built-in list keeps that property's field
and default (height starts at 100, color is a colour field). "object" is
reserved for the fixed input, checked by the command and on load.
The property picker names the owning type ("Sprite width"), and text
fields can now carry a suggestion list; Get/Set property use it.
Double-clicking a call node shows the path of opened behaviors under the
graph header. Clicking an entry jumps back and trims the path; choosing
a behavior any other way hides it. Reopening a behavior already on the
path steps back to it instead of repeating it. The bar shows before the
graph switches, otherwise it clips the graph that was just fitted.
The "Add node..." header button is gone; nodes come from the context menu.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Get/Set variable nodes now title themselves "Get <name>" / "Set <name>", and
the new Variable node just "<name>"; the title is the same folded or open,
and a collapsed node shrinks to it. The redundant "Variable: <name>" row is
gone unless the variable was deleted ("<name>?" is the only hint then).
The Variable node sets its value when an exec flow arrives and exposes the
value at any time. Its output is read when a downstream node pulls it, not
frozen at the write, so count = count + 1 is one Variable node plus an Add.
Get variable keeps reading once per pulse: switching it would change the
behaviour of existing games.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Measured click-to-paint was 72-77 ms, of which the editor itself spends
under 4 ms; the rest is the finished frame waiting in the compositor
queue. A desynchronized context bypasses it (16-18 ms in the same probe).
Pixi has no option for it, so main.ts builds the WebGL2 context itself.
The mode may tear on some drivers; ?vsync falls back to Pixi's own
context.
__gm.lowLatency() reports the attributes the browser actually granted,
and a graph-step check pins that the mode stays on.
Also adds the todo note graph-node-collapse-width.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Split GraphCanvas (1076 -> 831 lines) into EdgeLayer and graphMenus;
the port and marker menus were the same list and are now one function.
Value specs of the inspector and the list tab share valueSpecs.ts, and
the remaining copies (compound commands, type glyphs, typed asset
lookup, name rule, play/editor world setup, marquee) are merged.
Four copies had already drifted apart; merging them fixes:
- "Disconnect N links" in the port menu is one undo step, not N
- "+ Add collider" is hidden on a placed container instance, as in
the hierarchy menu, instead of always failing
- a bool property coerced from the text "false" becomes false
- a port command on an unknown node type reports instead of crashing
Comment-only edits leave the bundle byte-identical. README corrected in
35 places (step limit, jump-run clock, suite duration, 409 body, ...).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A chevron in front of the title toggles it without touching the selection,
and right-click offers Collapse node / Expand node for the whole selection
as one undo step. A collapsed node draws its title bar alone; every link of
one side meets at a single point on it, and hidden links there share one
mark per side, since per-port marks would stack on the same pixels. It has
no ports to grab, so a link dropped on it connects nothing.
Like hiding a link this is drawing only: collapsed is saved on the node and
an expanded node leaves no trace in the project file.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
buildStructBlock only covered structs, so a placed instance drew numbers,
text, bools and lists through createFieldRow on a 64 px name column while
its struct block sat on the asset's grid below it. variableRows.ts now
builds the row for every variable kind on both sides; the edit option
decides what the owner may change (asset: rename, x, item kind; instance:
revert and push). Label and Button take a fixed width so the mark column
no longer shifts names when a list shows its item-kind picker.
The two structFields calls that became dead with 9f48257 are gone. The
e2e steps locate the Enabled toggle through __gm.inspectorField instead
of clicking a fixed offset from its label.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Ein ausgeschaltetes Objekt zeichnet nichts, verschwand bisher aber auch aus
der Auswahl - im Panel angeklickt blieb der Viewport leer, es gab kein
Anzeichen, wo es liegt. Der Rahmen kommt jetzt, gedimmt auf 45 %, damit er
sich nicht wie eine fehlgeschlagene Zeichnung liest. Ziehen bleibt aus: der
Node ist weiterhin nicht interaktiv.
Dazu __gm.gizmos() als Testhaken - Gizmos waren vorher gar nicht prüfbar. Es
meldet das alpha, das wirklich in den Strich geht, kein daneben abgeleitetes
Flag.
Ein Toggle in der Graph-Kopfzeile lässt gezogene Nodes auf dem 24er-Raster
einrasten – derselben Größe, die den Hintergrund zeichnet. Einschalten bewegt
nichts; die Regel greift erst beim nächsten Zug. Bei Mehrfachauswahl richtet
sich jeder Node für sich aus, statt einem gemeinsamen Versatz zu folgen, der
die alten krummen Abstände mitschleppen würde.
Die Entscheidung liegt als reine Funktion in framing.ts, wo die übrige
Graph-Geometrie schon wohnt, und ist dort wie im Browser gepinnt.
A list item may now be a struct. The depth is bounded in the declaration
rather than by a comment: FieldKind (number|text|boolean|color) says what a
struct field holds, ListItemKind adds struct on top and says what a list
holds, so list -> struct -> primitive is the deepest storable shape and
sameValue stays finite. A struct item renders inline in the List tab, which
is why it needs no path in the slot address.
Four paths silently assumed a list item was a primitive: sameValue compared
items with ===, textOf would have printed [object Object], asListItem turned
a struct into text on Set item and Fill, and Index of used indexOf, which is
reference equality and could never have found a struct.
Tetris now runs on bundles. shapes is 112 {x, y} structs instead of 224 loose
numbers, piece and the candidate are structs, so Next move hands back a whole
candidate and each strand packs only the field it changes, since mergeStruct
leaves the rest standing. Declared behavior ports drop 49 -> 35, the master's
variables 18 -> 13, its graph 84/94 -> 69/76 nodes/edges.
Get variables and Set variables declare their ports from a list of variable
names, the way Pack and Expand already do, so one node can read or write
several at once. Renaming a variable rewires them, because there the port key
is the variable name.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A list property in the inspector was a single text field holding a comma
separated line -- unreadable past a handful of items, and it could not
carry anything but text. It becomes a button ("3 items") that opens a
sixth centre tab, where every item is a row built by the very widget the
inspector would build for that item kind.
The renderer was free: a PropertyDef of kind 'list' already carries
`items`, so item i becomes a synthetic def of that kind -- number, text,
boolean and colour rows appear without a new branch.
The way into the tab goes through the state (EditorState.listSlot,
openList(), event 'list'), like every other tab switch here; a tab may
not call showTab itself. ListSlot lives next to Selection rather than in
the tab folder, or state.ts would have to import upwards.
A row's handlers must not close over a snapshot of the list: they outlive
a commit, since the tab only rebuilds when the length changes, and the
second edited row would otherwise revive the value the first replaced.
showTab switched five `visible` lines by hand, where a sixth tab is
silently forgotten; it is a Record<CentreTab, Widget> now.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A rebuild destroys the widget being typed in, and a typed value can change the
very structure that decides whether a rebuild is needed -- a field that says
how many ports a node has. That made such a field unusable: every keystroke
previewed, the preview emitted 'project', refreshValues() saw the signature
change and threw the row away. Dragging the value worked only because a drag
keeps receiving globalpointermove on the destroyed widget.
So refreshValues() now postpones a due rebuild while the focus sits inside the
inspector, and runs it as soon as the field is left. Values are safe in the
meantime: TextField.setValue ignores an edit in progress.
Two handles for that in the toolkit. FocusManager.isWithin(node) also replaces
the parent chain refreshLive() walked by hand, and changeHandlers lets a panel
notice that the focus has gone -- there was no signal for that before.
Nothing in the editor can trigger this today, since the Sequence setting that
did is gone. It is the ground for the List tab, where a row that adds or
deletes an item changes the structure while you type. The e2e check therefore
builds the case out of real parts: type into a field, set the object's parent
from outside -- the parent is part of the structure signature -- and confirm
the input survives and the rebuild follows on Escape.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A held key fired once and then nothing, so sliding a Tetris piece sideways or
walking a menu meant building a timer next to every key node. Two settings on
the node do it now: Repeat after and Repeat interval, both in milliseconds so
the usual values stay whole numbers (300, 50) instead of 0.3 and 0.05. A delay
of 0 keeps the node exactly as it was, an interval of 0 means every frame.
The browser's own auto-repeat is deliberately not used - Input.press has always
dropped it - because its delay and rate are an operating-system setting of the
player's, not a game's. The repeat is built from the measured hold time instead,
per node, so two nodes on one key keep their own clocks. The rule itself is a
pure function in keyRepeat.ts and is checked without a world.
Both rows stay off the node body while the delay is 0: two lines of "0" on every
key node in every graph would be noise, and the screenshot said so. NodeDef gains
hiddenParams for that; the inspector still shows them.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A link can be folded away from the canvas: instead of a curve running clear
across the graph, each end grows a short stub with a mark naming the node at
the far side. The curve comes back while either of its two nodes is selected,
so following a connection is a click rather than a search, and a link carrying
a live pulse is always drawn so the trace keeps its thread.
Marks are collected per port, not per link, because several hidden links on one
port used to stack stubs, dots and labels on identical pixels. A bundle reads
"Move +1" and its dot selects every node the port feeds, as one undo step.
The whole thing is one optional flag on the edge; the interpreter never sees it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A behavior node that cannot run looked exactly like a healthy one: no
start node, no behavior picked, a deleted behavior or a self-call were
all silently skipped. The call still had to be forgiving at runtime -- a
dangling reference must not abort a game -- so the fix is diagnosis, not
an exception: describeProblem() names the one reason, the node draws it
in red under its title, and the inspector repeats it where you pick the
behavior. A missing end node is deliberately not a problem: a call is
always synchronous, and the end node only decides which values come back.
Two ways in: a behavior dragged from the asset panel lands in the graph
under the pointer, and double-clicking a call node opens the graph it
runs. Both reuse machinery that already existed -- the asset panel
already starts the drag, and list views already report double clicks.
The e2e module needed its assets at the head of the asset list: after 18
modules the list is long enough that a newly appended row scrolls out of
view, and the press then lands on the panel behind it, starting no drag
at all. That looked exactly like a rejected drop and cost two full runs.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A behavior that has a Start and an End node can now be dropped into any
other graph as a node of its own. What is declared on the start node
becomes the inputs of that node, what hangs on the end node its outputs;
the port list is derived from the sub graph, not copied, so declaring a
port in one graph grows the node in the other without touching it. The
node carries the behavior's name and the palette lists behaviors by
name, so a call is one grab instead of a node plus a dropdown.
An embedded behavior has no object and therefore no place for
variables: every call starts on the asset defaults and communicates
only through its ports. That is the conservative direction - state can
still be added later, and shared state already has a home in a behavior
object in the scene. Node ids are unique per graph only, so each call
gets its own pulse cache; without that, two graphs would collide.
Renaming a port used to silently drop the value typed on the calling
node. It is carried along now, project wide, like the edges.
Pause is a sub-state of play, not a third mode: the toolbar button and
Ctrl+Enter stop ticking the world, the clock included. A held frame still
drops the input edges, so resuming does not fire a pause worth of keys in
one go.
The editor run missed the case it is most wanted for. hasEditorRun only
looked at the stored scene objects, so a behavior inside a placed
container never started one; the world now builds from resolveObjects and
writes back through objectRef, keeping the move as an instance edit
instead of baking the instance out.
Keys and clicks go to the run while the Game tab is on screen, where no
key means anything to the editor, and are handed back on leaving it -
held keys released, so nothing keeps walking unpressed.
A behavior object carries a "Run in editor" switch next to Enabled. When
it is on, the behavior ticks outside of Play and works on the saved model
directly - Unity semantics: what it moves is a real change, autosaved and
not undoable.
Input stays with the editor, so On key never fires while typing. The
editor run never bakes placed containers into the model, since that would
silently drop their instanceEdits.
syncToModel reports whether it wrote anything, and the dirty flag is only
raised on change - otherwise a behavior moving something every frame
would trigger ~60 saves a second.
"+ → Texture" opened the machine's file dialog, so a picture that was uploaded once
but lost its asset – an undone import, a deleted asset – was unreachable from the
editor while still sitting on the server. Nothing collects those files, by design.
It now opens a grid of every picture the server holds, with thumbnails: pick one and
"Use image" makes a texture asset on that file, "Upload…" is the way in for bytes the
server does not have yet and lands in the same import as a drop. A tile captioned with
an asset name is one this project already wears – picking it selects that asset instead
of making a second one, since two assets on one file would fight over `filter`.
- `GET ?r=textures` now reports `width`/`height`; a texture asset holds the pixel size,
and reading it off the GPU texture would tie the asset's correctness to load timing.
- `TextureLibrary.load(file)` loads without touching `scaleMode` – a thumbnail must not
restage the filter of a picture the scene is drawing with.
- The grid geometry lives in `texture/imageGrid.ts` as pure functions; inside the widget
no test could reach it.
- The part after an upload is now one function shared by the drop import and the picker,
so "one file = one asset" is stated in a single place.
12 new E2E checks; the strongest one undoes an import, leaving the file on the server
with no asset naming it, and picks exactly that file back out of the dialog.
admin
