Decide USD collider visibility from purpose, and keep the render mesh when approximating by andrewkaufman · Pull Request #3712 · newton-physics/newton

andrewkaufman · 2026-07-28T19:22:29Z

Description

Decide what a USD collider draws from purpose and visibility, and stop physics:approximation from destroying the geometry it was only meant to collide with.

Three related defects, all in how parse_usd() assigns ShapeFlags.VISIBLE:

1. Drawability was inferred from a bound render material. A collider became visible only if it had PBR material cues (texture/roughness/metallic) — a guess at something USD already states declaratively. Because the supporting policy was model-wide, loading any visual geometry anywhere flipped unmaterialed colliders to hidden: a default-purpose prim serving as both visual and collider vanished because some other body authored a separate visual. Drawability now follows purpose resolving to default/proxy and the prim not being invisible.

2. physics:approximation replaced the render mesh. That attribute belongs to UsdPhysicsMeshCollisionAPI and is scoped to collision — it says what to collide against, not what to draw. The importer applied it destructively, so a prim that was both render geometry and a collider lost its authored topology and was drawn as its own collision hull. Approximating a drawable prim now splits it into an approximated collider plus a visual carrying the authored mesh, via the existing approximate_meshes(keep_visual_shapes=True).

3. An empty-viewport fallback overrode guide. When an import produced no visual shapes at all, every collider was given VISIBLE so the viewport would not come up blank. That overrode the one thing the asset had stated, drew guide geometry through the visual toggle, and — once approximated — drew the hull rather than the authored mesh. An asset whose geometry is all guide has no render geometry; an empty viewport is the correct depiction. force_show_colliders remains for inspecting such a scene.

Resulting contract

authoring	Newton shapes	Show Visuals	Show Colliders
`A(default)` + `B(guide, collider, approximated)`	2	A	B approximated
`A(default, collider, approximated)`	2	A authored topology	A approximated
`A(default, collider, not approximated)`	1	A	A
`A(guide, collider, approximated)`	1	nothing	A approximated
`A(guide, collider, not approximated)`	1	nothing	A

Row 3 stays a single shape deliberately: with no approximation the two would carry identical geometry, so splitting buys no independent control and guarantees z-fighting on coincident surfaces. The split earns its keep precisely when the geometries differ.

Closes #3650 (superseded by the general rule — a guide collider is not viewport geometry, so it is not drawn, and force_show_colliders still reveals it).

Effect on a real asset

A UR5e with a Robotiq 2F-85 gripper, identical camera, Show Collision off in both:

	`main`	this PR
stats overlay	`Shapes: 392 visual: 384 collision: 362`	`Shapes: 403 visual: 42 collision: 362`
gripper appearance	faceted hull fragments	authored meshes

Before:

After:

The gripper's link prims are dual-purpose — default purpose, PhysicsCollisionAPI applied, and physics:approximation authored (9 convexHull, 2 convexDecomposition). On main the approximation replaced their render geometry, so 11 prims became 353 hull fragments and those are what the visual toggle drew. Collider count is unchanged at 362 either way: this is purely what reaches the renderer, not what is simulated.

Worth noting the total shape count barely moves (392 → 403) while what is drawn goes 384 → 42. The totals alone do not show this, which is why the counts above are split (#3711).

Relationship to #3697

@eric-heiden diagnosed defect 1 independently in #3697 and reached the same predicate. This PR reaches the same result for shared geometry and additionally covers defects 2 and 3, which #3697 does not touch. Verified on the asset from that PR: the shared default-purpose collider is visible and the dedicated guide collider is not, identically on both branches. #3697's regression test is carried here unchanged, with credit — it pins the shared-geometry contract more explicitly than anything already in the suite.

#3694 makes defect 1 live on `main`

#3694 landed as e9aabe7d2, and it loads static visual-only geometry by default. Because the old policy was model-wide, more loaded visuals means more colliders silently hidden — so the failure is now reachable with ordinary scene content rather than requiring a second rigid body. On current main, adding a plain visual cube to a scene erases an unrelated dual-role collider:

                              main (e9aabe7d2)   this PR
static visual in scene=False  collider visible   collider visible
static visual in scene=True   *** NOTHING ***    collider visible

That is a table, a backdrop, or a floor removing a robot's shared collision/visual geometry from the viewport.

Checklist

New or existing tests cover these changes
The documentation is up to date with these changes
CHANGELOG.md has been updated (if user-facing change)

Test plan

uv run --extra dev -m unittest newton.tests.test_import_usd   # 288 tests
uv run --extra dev -m unittest newton.tests.test_model newton.tests.test_import_mjcf
uvx pre-commit run -a

Each behavioural commit was verified failing without its fix before being committed.

Swept every asset in newton/examples/assets and newton/tests/assets for shapes that stop being drawn. Exactly one changed: cartpole_single_pendulum.usda imported with nothing visible, because its three collision boxes are the cartpole's only geometry — they are its visuals as much as its colliders — yet were marked purpose="guide". The old fallback had been masking that. Fixed in the asset (final commit, three deleted lines) rather than by weakening the rule. No other bundled asset changes what it draws.

Bug fix

Steps to reproduce:

Author a rigid body whose single mesh prim is both render geometry and a collider — default purpose, PhysicsCollisionAPI applied.
Give it physics:approximation = "convexHull".
Import with ModelBuilder.add_usd().
The authored topology is gone: the prim imports as one shape whose geometry is the hull, and that hull is what the viewer draws as the object.

Separately, with any second body authoring a visual shape, a default-purpose collider with no bound material loses VISIBLE and is drawn by neither toggle.

Minimal reproduction:

import newton
from pxr import Usd, UsdGeom, UsdPhysics

stage = Usd.Stage.CreateInMemory()
UsdPhysics.Scene.Define(stage, "/physicsScene")
body = UsdGeom.Xform.Define(stage, "/Body")
UsdPhysics.RigidBodyAPI.Apply(body.GetPrim())

mesh = UsdGeom.Mesh.Define(stage, "/Body/Mesh")  # default purpose: render geometry
# ... author points/faceVertexIndices ...
UsdPhysics.CollisionAPI.Apply(mesh.GetPrim())
UsdPhysics.MeshCollisionAPI.Apply(mesh.GetPrim()).GetApproximationAttr().Set(
    UsdPhysics.Tokens.convexHull
)

builder = newton.ModelBuilder()
builder.add_usd(stage)

# Before: 1 shape, geometry replaced by the hull, drawn as the object.
# After:  2 shapes -- an approximated collider, and a visual with the authored mesh.
print(builder.shape_count)

Migration

Assets relying on the removed behaviours need one attribute:

A collider that should not be drawn now needs purpose = "guide". Previously a collider on a body with other visuals was hidden implicitly; that is no longer inferred.
A collision-only asset (all geometry guide) now imports with nothing visible. Pass force_show_colliders=True to inspect it.

hide_collision_shapes and force_show_colliders are unchanged.

🤖 Generated with Claude Code

Summary by CodeRabbit

Changed
- USD imports now determine collider visibility from USD purpose/visibility, with viewer override support and correct guide-purpose handling.
- Collider mesh approximation now preserves the authored visual topology for drawable approximations, while non-visual approximations no longer carry visual topology.
- Positive implicit Dahl-friction defaults are disabled unless explicitly authored (or compatibility mode is enabled).
Bug Fixes
- Prevented guide-purpose colliders with bound render materials from becoming visible unintentionally.
Tests
- Updated and expanded USD import/visibility and collider-vs-visual shape coverage.
Chores
- Updated the cartpole asset to remove guide purpose where applicable.

coderabbitai · 2026-07-28T19:22:48Z

📝 Walkthrough

Walkthrough

Changes

The USD importer now derives collider visibility from USD purpose and visibility, preserves authored visual topology when approximating viewport-drawn meshes, and updates shape flags and counts accordingly. Regression tests, an example asset, and changelog entries reflect these semantics.

USD import behavior

Layer / File(s)	Summary
Purpose-based collider visibility `newton/_src/utils/import_usd.py`, `newton/tests/test_import_usd.py`, `newton/examples/assets/cartpole_single_pendulum.usda`	Collider visibility follows USD purpose and importer overrides; guide-purpose colliders remain hidden unless explicitly forced visible.
Approximation and visual topology `newton/_src/utils/import_usd.py`, `newton/tests/test_import_usd.py`	Viewport-drawn approximated meshes receive separate visual copies retaining authored topology, while collision-only approximations omit visual topology.
Validation and release notes `newton/tests/test_import_usd.py`, `CHANGELOG.md`	Shape counts, flags, materials, mesh sources, and documented importer behavior are updated for the new results.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant USDStage
  participant USDImporter
  participant ModelBuilder
  USDStage->>USDImporter: provide purpose, visibility, and mesh data
  USDImporter->>ModelBuilder: create collider shapes and visual copies
  ModelBuilder->>ModelBuilder: apply visibility flags and collision approximation

Possibly related PRs

newton-physics/newton#3650 — Directly overlaps guide-purpose collider visibility with bound render materials.
newton-physics/newton#3697 — Overlaps shared collider/visual geometry and purpose-based visibility handling.
newton-physics/newton#3404 — Also changes guide-purpose visibility behavior in the USD importer.

Suggested reviewers: eric-heiden, adenzler-nvidia

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Out of Scope Changes check	⚠️ Warning	CHANGELOG.md also adds a SolverVBD friction-default update unrelated to the visibility and approximation scope in the linked issue and PR objectives.	Remove that changelog note or split the SolverVBD change into a separate PR unless its code change is included here.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	Guide-purpose colliders now stay off the visual toggle and can still be revealed with force_show_colliders, matching `#3650`.
Docstring Coverage	✅ Passed	Docstring coverage is 81.25% which is sufficient. The required threshold is 80.00%.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately captures the two main changes: collider visibility now follows USD purpose, and approximated meshes preserve render geometry.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

coderabbitai

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@CHANGELOG.md`:
- Line 47: Add migration guidance to the changelog entry describing the
additional visual shape created by ModelBuilder.add_usd() for approximated
renderable colliders. Explain that shape_count and shape-array iteration now
include this non-colliding visual copy, and instruct consumers to filter
collision shapes using ShapeFlags.COLLIDE_SHAPES or otherwise account for the
visual shape.

In `@newton/_src/utils/import_usd.py`:
- Around line 1210-1214: Update the documentation around the guide-purpose
handling to remove the obsolete material-inherited visibility explanation.
Describe guide as controlling viewport drawability, and state that
force_show_colliders is the explicit override while preserving the independently
evaluated collider display policy.
- Around line 3831-3832: Update the mesh-loading logic around _is_viewport_drawn
and the related visual-copy path to use _get_mesh_with_visual_material()
whenever load_visual_shapes is enabled and the prim is viewport-drawn. Ensure
synthesized visual meshes retain source texture, roughness, and metallic
metadata, including the default-purpose physics:approximation case with hidden
collision shapes.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yml

Review profile: CHILL

Plan: Pro Plus

Run ID: 0bc67b23-a382-436f-a4ed-f67a20f71756

📥 Commits

Reviewing files that changed from the base of the PR and between 6f1453b and d13b026.

📒 Files selected for processing (4)

CHANGELOG.md
newton/_src/utils/import_usd.py
newton/examples/assets/cartpole_single_pendulum.usda
newton/tests/test_import_usd.py

💤 Files with no reviewable changes (1)

newton/examples/assets/cartpole_single_pendulum.usda

coderabbitai

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

newton/tests/test_import_usd.py (1)
11250-11293: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Correct the stale default-visibility comment.

Line 11242 says this collider should not be visible, contradicting the new assertion on Line 11251. Update it to describe default-purpose collider visibility.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@newton/tests/test_import_usd.py` around lines 11250 - 11293, Update the stale
comment immediately before the default-purpose visibility assertions in the USD
import test to state that a default-purpose collider is visible when it is not
invisible. Keep the assertions and behavior unchanged.
Source: Path instructions

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In `@newton/tests/test_import_usd.py`:
- Around line 11250-11293: Update the stale comment immediately before the
default-purpose visibility assertions in the USD import test to state that a
default-purpose collider is visible when it is not invisible. Keep the
assertions and behavior unchanged.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yml

Review profile: CHILL

Plan: Pro Plus

Run ID: 485b9f93-a223-4cdb-b0fd-412fa377df4d

📥 Commits

Reviewing files that changed from the base of the PR and between d13b026 and 9c4b872.

📒 Files selected for processing (4)

CHANGELOG.md
newton/_src/utils/import_usd.py
newton/examples/assets/cartpole_single_pendulum.usda
newton/tests/test_import_usd.py

💤 Files with no reviewable changes (1)

newton/examples/assets/cartpole_single_pendulum.usda

🚧 Files skipped from review as they are similar to previous changes (2)

CHANGELOG.md
newton/_src/utils/import_usd.py

A collider with a bound render material was given the VISIBLE flag regardless of its purpose. Viewers draw a shape when (COLLIDE and show_collision) or (VISIBLE and show_visual), so such a collider was drawn by the visual toggle and could not be hidden by the collision toggle -- it occluded the very visual shapes it sat next to. "guide" is the conventional purpose for authored collision geometry, so honour it: purpose now gates only whether a collider inherits visibility from its render material. force_show_colliders is checked independently and still reveals guide colliders, which is what the previous behaviour was protecting. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

physics:approximation belongs to UsdPhysicsMeshCollisionAPI and is scoped to collision: it says which shape to collide against, not which to draw. The importer applied it destructively, so a Mesh that is both render geometry and a collider lost its authored topology and was drawn as its own collision hull. Approximating a prim that is viewport geometry now splits it into an approximated collider plus a visual carrying the authored mesh, via approximate_meshes(keep_visual_shapes=True). Whether a prim is viewport geometry follows USD purpose and visibility alone: an unauthored purpose composes to "default" and is drawable, and no bound material is required. The collider display policy that governs pure colliders does not decide this, since such a prim is render geometry in its own right. Colliders with no approximation are left alone -- their collision geometry is their authored geometry, so there is nothing to preserve -- as are guide-purpose and invisible prims, which have no render role. The visual carries zero density and no collision, and copies append after the originals, so masses, existing shape indices and path_shape_map entries are unchanged. Both spellings of the same intent now import alike: a guide-purpose collider beside a visual mesh, or one dual-purpose prim, each yielding one collider and one visual. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Whether a collider is drawn was inferred from whether a render material with PBR cues happened to be bound to it. USD already states this declaratively: a prim whose purpose resolves to "default" or "proxy" and which is not invisible is viewport geometry, whether that purpose was authored or inherited from the fallback, and regardless of any material. The heuristic was not merely redundant, it was wrong in a way that got worse with more visuals in a scene. Because the collider display policy is model-wide, loading any visual geometry anywhere flipped unmaterialed colliders to hidden. With static visual shapes now imported by default, dropping a visual cube into a scene erased an unrelated collider entirely -- it rendered nothing at all. This subsumes the guide-purpose fix in the preceding commit by the more general rule: a guide collider is not viewport geometry, so it is not drawn, and force_show_colliders still reveals it. Two tests asserted the heuristic and are updated to the purpose-based contract, including the guide-purpose escape hatch that replaces the behaviour they were pinning. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Collider visibility had a fallback: when an import produced no visual shapes at all, every collider was given VISIBLE so the viewport would not come up empty. That overrode the one thing the asset had actually stated. A prim marked guide is not viewport geometry, and a scene whose geometry is all guide has no render geometry -- an empty viewport is the correct depiction of it, not a problem to paper over. The fallback also drew such colliders through the visual toggle, so Show Collision could not hide them and, once approximated, what it drew was the hull rather than the authored mesh. Visibility now follows the asset: purpose and visibility decide, with force_show_colliders as the explicit override for inspecting a collision-only scene. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

These three boxes are the cartpole's only geometry -- they are its visuals as much as its colliders -- so marking them guide asserted something untrue about the asset. The importer previously compensated with a fallback that showed colliders when a scene had no visual shapes, which hid the mistake; without it the asset imported with nothing visible at all. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

@eric-heiden

Port the regression test from @eric-heiden's PR newton-physics#3697, which found the same bug independently: one default-purpose prim serving as both visual and collider lost VISIBLE as soon as another body authored separate visual geometry. Passes here unchanged, and pins the shared-geometry contract more explicitly than the existing cases did. Co-Authored-By: Eric Heiden <eric-heiden@users.noreply.github.com> Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

The empty-viewport fallback removal had no regression test: the existing collision-only case authors a default-purpose sphere, so it exercises drawability rather than the fallback. Pin the case that actually changed -- every prim guide, nothing drawn, force_show_colliders to inspect. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Load the render material for a mesh whenever something will draw it, not only when the collider itself is visible. hide_collision_shapes clears collider_is_visible while the visual copy is still produced, so that copy could be built from a mesh loaded without texture, roughness or metallic data. Update the purpose predicate's docstring, which still described gating material-inherited visibility after that heuristic was removed, and give the changelog entry migration guidance for the extra shape. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

andrewkaufman · 2026-07-28T19:52:17Z

Rebased onto e9aabe7d2 now that #3694 has landed.

That merge changes this from a latent bug to a live one. #3694 loads static visual-only geometry by default, and because the old collider-visibility policy was model-wide, any loaded visual flips unmaterialed colliders to hidden. On current main, dropping a plain visual cube into a scene now erases an unrelated dual-role collider outright:

                              main (e9aabe7d2)   this PR
static visual in scene=False  collider visible   collider visible
static visual in scene=True   *** NOTHING ***    collider visible

In practice that is a table, floor, or backdrop removing a robot's shared collision/visual geometry from the viewport. Previously it took a second rigid body authoring a visual to trigger; now ordinary scene content does it.

292 tests pass on the rebase, and all five authoring cases in the table above still hold.

coderabbitai

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

newton/_src/utils/import_usd.py (1)

3298-3302: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win

Avoid treating all static geometry as one visual body.

Loading any static visual records body -1; then every static collider matches it and is hidden by hide_collision_shapes=True, even when unrelated or force-shown.

Proposed fix

-                has_body_visual_shapes = load_visual_shapes and body_id in bodies_with_visual_shapes
+                has_body_visual_shapes = load_visual_shapes and body_id >= 0 and body_id in bodies_with_visual_shapes

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@newton/_src/utils/import_usd.py` around lines 3298 - 3302, Update the
static-visual loading loop to avoid passing the shared body ID -1 to
_load_visual_shapes_impl for every prim. Assign each unrelated static visual its
own identity or otherwise prevent it from matching all static colliders, while
preserving exclusions for deformable_visual_exclude_paths, path_shape_map, and
rigid-body hierarchies.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In `@newton/_src/utils/import_usd.py`:
- Around line 3298-3302: Update the static-visual loading loop to avoid passing
the shared body ID -1 to _load_visual_shapes_impl for every prim. Assign each
unrelated static visual its own identity or otherwise prevent it from matching
all static colliders, while preserving exclusions for
deformable_visual_exclude_paths, path_shape_map, and rigid-body hierarchies.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yml

Review profile: CHILL

Plan: Pro Plus

Run ID: 41da0d2f-4a7d-4e5a-8829-d35d69a30e25

📥 Commits

Reviewing files that changed from the base of the PR and between 9c4b872 and d8625bb.

📒 Files selected for processing (4)

CHANGELOG.md
newton/_src/utils/import_usd.py
newton/examples/assets/cartpole_single_pendulum.usda
newton/tests/test_import_usd.py

💤 Files with no reviewable changes (1)

newton/examples/assets/cartpole_single_pendulum.usda

🚧 Files skipped from review as they are similar to previous changes (2)

CHANGELOG.md
newton/tests/test_import_usd.py

codecov · 2026-07-28T20:11:22Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!

github-actions Bot assigned andrewkaufman Jul 28, 2026

andrewkaufman mentioned this pull request Jul 28, 2026

Add load_usd example for loading arbitrary USD files #3629

Draft

3 tasks

coderabbitai Bot reviewed Jul 28, 2026

View reviewed changes

Comment thread CHANGELOG.md Outdated

Comment thread newton/_src/utils/import_usd.py Outdated

Comment thread newton/_src/utils/import_usd.py

andrewkaufman force-pushed the fix-shape-loading branch from d13b026 to 9c4b872 Compare July 28, 2026 19:32

andrewkaufman mentioned this pull request Jul 28, 2026

Fix shared USD collider visibility #3697

Closed

3 tasks

coderabbitai Bot reviewed Jul 28, 2026

View reviewed changes

andrewkaufman requested a review from eric-heiden July 28, 2026 19:41

StafaH approved these changes Jul 28, 2026

View reviewed changes

Comment thread newton/_src/utils/import_usd.py

andrewkaufman and others added 8 commits July 28, 2026 12:49

andrewkaufman force-pushed the fix-shape-loading branch from 9c4b872 to d8625bb Compare July 28, 2026 19:50

coderabbitai Bot reviewed Jul 28, 2026

View reviewed changes

eric-heiden approved these changes Jul 28, 2026

View reviewed changes

eric-heiden added this pull request to the merge queue Jul 28, 2026

Merged via the queue into newton-physics:main with commit 92ae945 Jul 28, 2026
30 checks passed

andrewkaufman deleted the fix-shape-loading branch July 28, 2026 20:58

coderabbitai Bot mentioned this pull request Jul 29, 2026

[Kamino] Fix Kamino USD importer to align with Newton importer #3721

Merged

3 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Decide USD collider visibility from purpose, and keep the render mesh when approximating - #3712

Decide USD collider visibility from purpose, and keep the render mesh when approximating#3712
eric-heiden merged 8 commits into
newton-physics:mainfrom
andrewkaufman:fix-shape-loading

andrewkaufman commented Jul 28, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Jul 28, 2026 •

edited

Loading

Walkthrough

Changes

Sequence Diagram(s)

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

andrewkaufman commented Jul 28, 2026

Uh oh!

coderabbitai Bot left a comment

Uh oh!

codecov Bot commented Jul 28, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

andrewkaufman commented Jul 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Resulting contract

Effect on a real asset

Relationship to #3697

#3694 makes defect 1 live on main

Checklist

Test plan

Bug fix

Migration

Summary by CodeRabbit

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Jul 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Sequence Diagram(s)

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

andrewkaufman commented Jul 28, 2026

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

codecov Bot commented Jul 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

andrewkaufman commented Jul 28, 2026 •

edited

Loading

#3694 makes defect 1 live on `main`

coderabbitai Bot commented Jul 28, 2026 •

edited

Loading

codecov Bot commented Jul 28, 2026 •

edited

Loading