Commit Graph

387 Commits

Author SHA1 Message Date
Keavon Chambers
4f49c000e8 Rename the nodes 'Length' -> 'Magnitude', 'Flatten Path' -> 'Combine Paths', 'Vec2 Value' -> 'Combine Vec2', and add a new 'Vec2 Value' node (#4349)
* Rename the node 'Length' -> 'Magnitude'

* Rename the node 'Flatten Path' -> 'Combine Paths'

* Replace the node 'Vec2 Value' with 'Combine Vec2' and add a new 'Vec2 Value' that's actually a vec2

* Update demo artwork
2026-09-15 14:35:48 +02:00
Keavon Chambers
04d6c0d5cf Convert the node catalog to rank-polymorphic kernels, materialize stored values as ranked wires, and display wire rank in the graph
Co-authored-by:    Dennis Kobert <dennis@kobert.dev>
2026-09-15 14:35:20 +02:00
Keavon Chambers
9dd1e2bdf9 Rename GradientStops to Gradient and the legacy Gradient/Fill structs to LegacyGradient/LegacyFill 2026-09-15 14:34:59 +02:00
Dennis Kobert
3bdec5d005 Fold attribute names through monitors and skip coercing name inputs 2026-09-15 14:33:45 +02:00
Dennis Kobert
5baeb8fd46 Accept any record wire on an input read only for its attributes
An element generic declared on a reading input, absent from the output
and from every other input, has nowhere to go and nothing to be: the
node reads its declared attributes and never the element. That is an
attr-to-element map, so it accepts any upstream record wire.

The blessing is inferred rather than marked, since the signature already
says it. The two substrate pieces were both in place: the element rides
as the byte-carried token, so the registry takes one generic row instead
of a row per element type, and the reads resolve against the input's own
layout as they always have. Only the passthrough rule stood in the way,
in the three places that enforced it.

A generic the output carries, or one on an input with no reads, is
unchanged and still owes an implementations list.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-09 12:15:40 +00:00
Dennis Kobert
f04d439b33 Name the attribute read for the type it reads
The read is a number read, not a read of whatever the name holds: our
model has no coercion, so a name written at another value type is a graph
error rather than a conversion. The name says so.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-09 11:35:38 +00:00
Dennis Kobert
6e22d6a412 Serve a declared name's census default on an absent read
An absent read of a name the census declares now serves that name's own
default rather than its value type's, which is the rule for a declared
name. The census stages it exactly as it fills any absent field, as the
declared default's bytes, so the read reuses the census's own mechanism
instead of a second one; the compiler takes them when the name folds and
`set_layout` installs them beside the offset.

A name the census does not declare keeps the value type's default, which
is that case's own rule. One name carries one value type, checked when
the name folds, so the row's value type and width agree with the read's.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-09 11:13:41 +00:00
Dennis Kobert
5eb80721dd Resolve a named read's offset when the graph compiles
The fold already holds the name and the read input's finished layout, so
the offset falls out there rather than at construction: `RecordLayout`
carries the resolved numbers and `set_layout` copies them into the read
slots. Constructors are untouched, and census-marker reads keep their
current installation.

A read meets the value type the name was written at, so a disagreement
between a read here and a write upstream is the same graph error as two
writes disagreeing; the one-name-one-type check now spans reads and
writes together. An absent attribute stays absent and the read serves
the forced default rather than reporting it.

`read_attribute` is the catalog's get half, typed and never `Option` at
the kernel boundary, with the name declared exactly as the write side
declares it.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-09 10:50:08 +00:00
Dennis Kobert
9b92edc0e9 Write an attribute under a name the graph supplies
`Named<X>` in parameter position declares where a placeholder's name is
wired, and the macro gives that input constant text; `Attr<Named<X, V>>`
in the return writes under it. The kernel is handed the bare placeholder,
since the name is spent resolving the layout and a folded offset is all
the write needs, so the hot path matches a marker node's exactly.

A name-generic write names its value type through the wired generic,
which only an implementations row resolves, so those nodes emit their
layout meta per row rather than sharing one across rows.

`write_attribute` is the catalog's set half, restoring the identifier
master's documents carry with its input positions. Its name input is a
constant, so those documents resolve without migration, and the reset
marker that stood in for the missing node retires with it.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-09 09:26:17 +00:00
Dennis Kobert
348e9022a6 Fold attribute names out of their constant inputs when the graph compiles
`compute_layouts` resolves every name-from-input write against the text
its name input carries, so a folded meta is indistinguishable from a
marker node's and the layout holds a `&'static str` from there on. That
leaves nowhere for a name to be computed, which is the whole of the
rule: a name input that is not a constant is refused right here, per
node, with the path the editor pins its diagnostic to.

One name carries one value type, checked across the census and the
names a node folds together, so no graph can declare one field at two
widths. `compute_layouts` returns those refusals rather than panicking.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-09-09 08:51:05 +00:00
Dennis Kobert
b321c094d2 Keep the new assert messages inside the format width 2026-09-08 13:29:09 +00:00
Dennis Kobert
5e388cb38f Assert the deep element glue is registered before the executor evaluates 2026-09-08 13:15:06 +00:00
Dennis Kobert
33f7257e47 Rename record_edge_type to record_source_type 2026-09-06 16:27:11 +00:00
Dennis Kobert
daf0231c0c Rename the record_edge accessor to record_source 2026-09-06 16:26:33 +00:00
Dennis Kobert
92adc524ca Rename EdgeHandle to SourceHandle 2026-09-06 16:26:03 +00:00
Dennis Kobert
9f0e0ebc96 Clear the remaining clippy warnings on the diff surface 2026-09-06 16:25:17 +00:00
Dennis Kobert
e92faf763d Correct docs that misstate what the code does 2026-09-06 15:58:01 +00:00
Dennis Kobert
9d3abe385f Sweep the data-flow wire and edge wording to input and source 2026-09-06 15:14:55 +00:00
Dennis Kobert
8f7d3936f8 Rename serve_edge to serve_input 2026-09-06 15:12:46 +00:00
Dennis Kobert
c12f8f1115 Add changing-seasons to the gungraun benchmark arguments 2026-09-01 13:18:01 +00:00
Dennis Kobert
4e7d72d4b8 Flush the persistent region on its heap budget 2026-08-31 23:52:44 +00:00
Dennis Kobert
5a6192bbd6 Give the executor a persistent arena flushed between evaluations 2026-08-29 22:07:27 +00:00
Dennis Kobert
a955d6b458 Run cargo fmt 2026-08-29 18:40:02 +00:00
Dennis Kobert
97bc10e4c3 Pass frame space by argument and delete the record stack 2026-08-29 18:33:23 +00:00
Dennis Kobert
3af6834d3c Flip the Node trait from eval onto the frame claim's serve 2026-08-29 12:26:46 +00:00
Dennis Kobert
277641d27e Delete the plain edge kind 2026-08-29 10:07:10 +00:00
Dennis Kobert
f4d37524c8 Structure record stack rewinds as scope guards and make reserve unsafe 2026-08-28 18:02:50 +00:00
Dennis Kobert
50911a6a25 Recreate introspection data from the monitor's context snapshot 2026-08-27 13:53:25 +00:00
Dennis Kobert
20e70e7bc9 Exclude executor teardown from the measured bench regions 2026-08-27 10:09:18 +00:00
Dennis Kobert
694d3d34b3 Reduce measurement overhead for run_once benchmark 2026-08-27 00:37:25 +02:00
Dennis Kobert
dcc9059041 Give To Graphic a unit row so an unconnected content input renders as nothing 2026-08-24 12:02:46 +00:00
Dennis Kobert
aba941e81d Collapse typed levels in To Graphic like the pre-flip list conversion 2026-08-24 01:35:32 +00:00
Dennis Kobert
c0eb2ee4ff Remove the runtime-named attribute family and its type-erased bridges 2026-08-23 19:05:26 +00:00
Dennis Kobert
94b9d279a2 Retarget the palette executor test at the folded record wire 2026-08-23 17:29:43 +00:00
Dennis Kobert
5e918af371 Restore the sample_polyline test attribute and drop dead scaffolding 2026-08-23 08:31:40 +00:00
Dennis Kobert
58182bc2d3 Keep the executor arena at 128 MiB 2026-08-23 00:57:37 +00:00
Dennis Kobert
c33f8ae75b Convert mirror to a fold emitter and bridge levels into the erased list 2026-08-22 22:15:03 +00:00
Dennis Kobert
3126a5b278 Convert copy to points, sample polyline, scatter, spline, jitter, and assign colors 2026-08-22 20:59:37 +00:00
Dennis Kobert
5f6c18893e Grow the executor arena for per-copy paint parks 2026-08-22 20:59:37 +00:00
Dennis Kobert
6ab9d0360a Convert flatten path and solidify stroke to fold kernels 2026-08-22 19:11:48 +00:00
Dennis Kobert
01da817f61 Grow the executor arena for fold snapshots 2026-08-22 19:11:48 +00:00
Dennis Kobert
3da7f6cea3 Drop the legacy fill and stroke rows for single-typed paint 2026-08-22 17:29:51 +00:00
Dennis Kobert
4bceae26f2 Add the leveled fill rows and the graphic coercion entries 2026-08-22 16:13:01 +00:00
Dennis Kobert
0b2aa3798a Derive extent forwarding for level-preserving passthroughs 2026-08-22 15:34:22 +00:00
Dennis Kobert
fb71ea426d Bridge legacy seams and convert the transform family 2026-08-22 14:25:31 +00:00
Dennis Kobert
7fb36123e9 Point the render wrapper at the leveled intermediate 2026-08-22 12:50:47 +00:00
Dennis Kobert
5821816228 Add the leveled boundary helpers and the deep group element copy 2026-08-22 11:48:26 +00:00
Dennis Kobert
fbba0cd037 Gate the gpu converter imports the sweep wrongly removed 2026-08-17 10:56:45 +00:00
Dennis Kobert
ae7c51a964 Drop stale imports left behind by the record-wire port 2026-08-17 10:28:29 +00:00
Dennis Kobert
99ce2ea565 Require a node registry to compile so the layout pass always runs 2026-08-15 14:49:25 +00:00