cuda-core experimental namespace deprecation (attempt v0) #1298

rwgk · 2025-12-02T00:35:19Z

This was the initial attempt — continuing the work under #1377 (starting over fresh from main)

[UNTESTED] This commit moves all implementation files from the experimental namespace to the core namespace. The experimental/ directory is kept for backward compatibility stubs (to be added in a later commit). Files moved: - All .py, .pyx, .pxd files from experimental/ to core/ - _memory/, _utils/, include/ subdirectories - __init__.pxd renamed to __init__experimental.pxd This is a file move only - imports will be updated in the next commit.

[UNTESTED] This commit updates all import statements in the moved files to reference the new cuda.core.* paths instead of cuda.core.experimental.*. Updated: - Python files (.py): _module.py, _graph.py, _linker.py, _program.py, _system.py, utils.py - Cython files (.pyx): _device.pyx, _stream.pyx, _launcher.pyx, _launch_config.pyx, _event.pyx - Memory subdirectory: all .py, .pyx, and .pxd files - Utils subdirectory: cuda_utils.pyx - Other files: _kernel_arg_handler.pyx, _memoryview.pyx All imports now use the new non-experimental paths.

[UNTESTED] This commit updates cuda/core/__init__.py to export all the symbols that were moved from experimental namespace. The imports now reference the new cuda.core.* paths instead of cuda.core.experimental.*. Exported symbols: - Device, Event, EventOptions - Graph, GraphBuilder, GraphCompleteOptions, GraphDebugPrintOptions - LaunchConfig, launch - Linker, LinkerOptions - Memory classes: Buffer, DeviceMemoryResource, etc. - Kernel, ObjectCode - Program, ProgramOptions - Stream, StreamOptions - System, system - utils module

…warnings [UNTESTED] This commit creates backward compatibility stubs in the experimental namespace that forward imports to the new cuda.core.* locations and emit deprecation warnings. Features: - All public symbols are re-exported from new locations - DeprecationWarning is emitted when the experimental module is imported - __getattr__ provides lazy forwarding for submodules - Clear migration message in docstring The stubs ensure backward compatibility while encouraging users to migrate to the new import paths.

[UNTESTED] This commit updates all test files to import from the new cuda.core.* paths instead of cuda.core.experimental.*. Updated 28 test files: - All test_*.py files - Helper files in tests/helpers/ - Memory IPC test files - Example test files Tests now exercise the new API paths directly, ensuring the migration is complete. Backward compatibility will be tested separately.

[UNTESTED] Update the remaining .pyx test file to use cuda.core.* imports.

[UNTESTED] This commit adds comprehensive tests for the experimental namespace backward compatibility stubs. Tests verify: - Experimental imports still work and emit deprecation warnings - Symbols from experimental are the same objects as from core - Direct imports from experimental work correctly - Utils module and options classes are accessible - Memory classes and other public APIs work - Objects can be instantiated through experimental namespace These tests ensure backward compatibility while the migration happens.

[UNTESTED] Update all example files to import from cuda.core instead of cuda.core.experimental. Examples should demonstrate the new recommended import paths.

copy-pr-bot · 2025-12-02T00:35:22Z

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[UNTESTED] Update build_hooks.py to look for Cython extensions in cuda/core/ instead of cuda/core/experimental/. This fixes the build system to work with the moved files.

[UNTESTED] Update _context.pyx to import from cuda.core._utils.cuda_utils instead of cuda.core.experimental._utils.cuda_utils. This fixes a circular import issue during module initialization.

[UNTESTED] Update _memory/_legacy.py to import from cuda.core._utils.cuda_utils instead of cuda.core.experimental._utils.cuda_utils. This fixes a circular import issue.

[UNTESTED] Update _memory/_virtual_memory_resource.py to import from cuda.core._utils.cuda_utils instead of cuda.core.experimental._utils.cuda_utils. This fixes a circular import issue.

[UNTESTED] Update cuda_python_test_helpers to import from cuda.core._utils instead of cuda.core.experimental._utils.

[UNTESTED] Update test files to access internal modules via cuda.core._* instead of cuda.core.experimental._*. This fixes test failures after the namespace migration.

[UNTESTED] Make utils accessible as cuda.core.experimental.utils submodule for backward compatibility. Also fix test_device_id to use cuda.core.system instead of cuda.core.experimental.system.

[UNTESTED] Clear cached experimental module before testing to ensure warnings are emitted. Also improve assertion message.

kkraus14 · 2025-12-02T01:49:26Z

cuda_core/tests/test_experimental_backward_compat.py

+
+        # Should emit deprecation warning
+        assert len(w) >= 1
+        assert issubclass(w[0].category, DeprecationWarning)


We should be able to leverage with pytest.deprecated_call https://docs.pytest.org/en/stable/reference/reference.html#pytest.deprecated_call

cursor-agent:

Good catch! I've updated the tests to use pytest.deprecated_call() instead of manual warning catching. This is cleaner and more idiomatic pytest usage.

human:

I'll work in this auto-pilot mode until I get everything passing here (CI). Then I'll review myself.

kkraus14 · 2025-12-02T01:54:14Z

cuda_core/cuda/core/experimental/__init__.py

+                "_linker", "_memory", "_module", "_program", "_stream", "_system", 
+                "_utils", "_context", "_dlpack", "_kernel_arg_handler", 
+                "_launch_config", "_memoryview"):
+        _warn_deprecated()


Does this trigger a warning on each call as opposed to just at import time?

cursor-agent:

The deprecation warning is only emitted once at import time when cuda.core.experimental is first imported, not on each attribute access. This is because:

The warning is emitted at module initialization (line 64: _warn_deprecated())

utils is already set as an attribute in the module namespace (line 68), so accessing it doesn't go through __getattr__

Only accessing submodules via __getattr__ (like _device, _utils) triggers additional warnings

I've added documentation to clarify this behavior.

The call to _warn_deprecated in this function will never cause a warning to be issued because the warning was already issued at module scope when loading this module. It should be removed.

Do we really intend to publish all submodules by stripping the leading underscores? That would make public many things that are currently private.

This implementation pays for the extra attribute look-up with each access. What about something like the following?

# Fallback to underscore-prefixed name attr = importlib.import_module(f"mymodule.{name}") # Set the attribute globally to avoid __getattr__ next time setattr(sys.modules[__name__], name, attr) # alternative: globals()[name] = attr return attr

However, at that point, it might be better to just say this:

from .. import _device, _event, # etc.

I prefer strongly that we just remove this entirely. Underscored modules are never considered public APIs.

This merge brings in the latest changes from main (commit a4b285a) including the peer access control for DeviceMemoryResource (NVIDIA#1289). The merge combines: - Latest main changes (a4b285a) - Experimental namespace deprecation work from this branch All files remain moved from cuda.core.experimental.* to cuda.core.* with backward compatibility stubs maintained.

rwgk · 2025-12-02T04:52:14Z

Remark: I haven't looked at any of the changes myself yet. Everything to this point was done by cursor-agent. I just gave it a few prompts to explain what I want to do.

…ng behavior - Replace manual warning catching with pytest.deprecated_call() for cleaner tests - Add documentation clarifying that deprecation warnings are only emitted at module import time, not on each attribute access - Update test_experimental_utils_module docstring to explain warning behavior

Update import from cuda.core.experimental._utils.cuda_utils to cuda.core._utils.cuda_utils to match the new module structure.

rwgk · 2025-12-02T05:11:47Z

/ok to test

- Update cython/build_tests.sh to use cuda/core/include instead of cuda/core/experimental/include - Update test_memory_peer_access.py to import from cuda.core instead of cuda.core.experimental - Update test_module.py to use cuda.core.LaunchConfig instead of cuda.core.experimental.LaunchConfig

rwgk · 2025-12-02T06:00:59Z

/ok to test

github-actions · 2025-12-02T06:10:21Z

Doc Preview CI
🚀 View preview at https://nvidia.github.io/cuda-python/pr-preview/pr-1298/
https://nvidia.github.io/cuda-python/pr-preview/pr-1298/cuda-core/
https://nvidia.github.io/cuda-python/pr-preview/pr-1298/cuda-bindings/
https://nvidia.github.io/cuda-python/pr-preview/pr-1298/cuda-pathfinder/
Preview will be ready when the GitHub Pages deployment is complete.

Andy-Jost · 2025-12-03T16:16:32Z

Another approach would be to change absolute imports to relative, e.g.:

- from cuda.core.experimental._utils.cuda_utils import driver
+ from ._utils.cuda_utils import driver

An advantage of this is that it decouples the code changes from the file movements, so it could be checked into the current main without eliminating the experimental namespace. After that, the files could be moved without changing any import statements.

I've been planning to propose this change but haven't had time to ask the team for input yet.

Resolved conflicts by updating imports from cuda.core.experimental.* to cuda.core.* to match the branch's deprecation work. Updated all source files and test files to use the new import paths.

rwgk · 2025-12-09T19:30:55Z

/ok to test

rwgk · 2025-12-09T20:29:02Z

Regarding absolute vs relative imports:

I've been following the direct guidance of @kkraus14, which he gave me/us around the time when I started working on cuda-pathfinder (~spring this year). It was that we should use absolute imports throughout.

For things like this, I usually put my personal preference aside completely. There will always be valid arguments one way or the other. In the end what I think is most valuable is consistency, ideally as enforced by tooling (with overrides that must be explicit, e.g. noqa).

I'd be a big fan of adding TID252 to our ruff.toml, and then agree that everything that passes that check is fine.

cuda_core/cuda/core/experimental/__init__.py

leofang · 2025-12-09T20:42:44Z

cuda_core/cuda/core/include/utility.hpp

We probably need to document that everything under cuda_core/cuda/core/include/ are considered internal (for now). There is currently no plan to maintain any non-Python (such as C, C++, and Cython) public APIs in cuda-core, and when we do we will revise this document.

Resolved conflicts by updating imports from cuda.core.experimental.* to cuda.core.* to match the branch's deprecation work. Updated all source files and test files to use the new import paths. Added new StridedLayout imports from experimental namespace as they are new features from main that haven't been migrated yet.

rwgk · 2025-12-10T23:15:41Z

/ok to test

cuda_core/cuda/core/_memory/_buffer.pyx

cuda_core/cuda/core/experimental/__init__.py

Merged latest changes from main including: - Added size property to StridedMemoryView (NVIDIA#1363) - NVML bindings updates (NVIDIA#1360) - cuda_pathfinder pixi configuration (NVIDIA#1365, NVIDIA#1361) All changes merged cleanly with no conflicts.

- Moved _layout.pxd and _layout.pyx from experimental/ to cuda/core/ - Updated all imports from cuda.core.experimental._layout to cuda.core._layout - Added StridedLayout to cuda.core.__init__.py exports - Added backward compatibility stub in experimental/__init__.py - Fixed docstring references in _linker.py and _module.py - Updated pyproject.toml package-data path - Updated docs conf.py excluded_dirs All code has been migrated from cuda.core.experimental except backward compatibility stubs.

…launch_config).

The TYPE_CHECKING block was unnecessary since all symbols are already imported at runtime below. Type checkers will see the types from the runtime imports, making the TYPE_CHECKING block redundant duplication.

- Update issue templates to use cuda.core.Program instead of cuda.core.experimental.Program - Update merge script to place versioned directories (cu12, cu13) under cuda/core/ instead of cuda/core/experimental/ - Fix duplicate line in merge script

rwgk · 2025-12-15T17:30:22Z

/ok to test

The merge script was copying the entire cuda/core directory into versioned subdirectories and then removing all Python modules from cuda/core, causing ModuleNotFoundError when importing cuda.core. Fix: Only copy version-specific binaries (.so, .pyd, .dll) into versioned directories (cu12/, cu13/), and keep all Python modules in cuda/core/ where they belong. Since wheels no longer have experimental/ directory, we can simplify the merge logic compared to the original approach which assumed experimental/ contained only binaries.

rwgk · 2025-12-16T03:28:45Z

/ok to test

The rglob("*") was recursively searching through all subdirectories, including the cu12/ and cu13/ directories we just created. This caused files from cuda/core/cu13/ to be copied to cuda/core/cu13/cu13/, creating infinite recursion. Fix: Skip any files that are in versioned directories (cu12/, cu13/) when searching for binaries to copy.

rwgk · 2025-12-16T04:44:59Z

/ok to test

leofang · 2025-12-16T05:00:08Z

cuda_core/cuda/core/__init__.py

+from cuda.core._memory import (  # noqa: E402
+    Buffer,
+    DeviceMemoryResource,
+    DeviceMemoryResourceOptions,
+    GraphMemoryResource,
+    LegacyPinnedMemoryResource,
+    MemoryResource,
+    VirtualMemoryResource,
+    VirtualMemoryResourceOptions,
+)


This list needs to be updated to capture a few new things.

leofang · 2025-12-16T05:01:14Z

cuda_core/cuda/core/__init__experimental.pxd

Cython needs an exact filename __init__.pxd so as to find the current modules. This renaming does not make sense to me. What does it do?

Probably an accident.

I had to git merge main already several times, a likely cause of accidents. The merges are messy. I'm using Cursor and it seems to miss some details sometimes.

This PR has become much more complex than it looked like initially (tests passed a couple weeks ago).

E.g. I don't understand why ci/tools/merge_cuda_core_wheels.py (last changed in Oct) worked before but now needs a change (it makes sense that it needs a change). Testing it locally isn't easy because the input wheels only live on the runners and disappear when the jobs finish. So I'd have to figure out how to build them locally (currently I'm not trying to).

I'll keep beating on it until I'll get all tests to pass again. Maybe I'll start over fresh. LLMs seems to be less successful when working incrementally.

Thanks for the hints.

Ah! Leaving a quick note before signing off...

The merge script can be tested locally easily. Create two conda envs, one with CUDA 13 and another with CUDA 12, and build cuda.core wheels in each env via pip wheel --no-deps --no-build-isolation -v .. The cuda-bindings package can just be conda-installed together with the CTK for simplicity. No need to build it from source. Then, we take the two wheels and rename them before running the merge script. This logic is relevant:

cuda-python/.github/workflows/build-wheel.yml

Lines 430 to 439 in 83eaec1

# Rename wheel to include CUDA version suffix

mkdir -p "${{ env.CUDA_CORE_ARTIFACTS_DIR }}/cu${BUILD_PREV_CUDA_MAJOR}"

for wheel in ${{ env.CUDA_CORE_ARTIFACTS_DIR }}/*.whl; do

if [[ -f "${wheel}" ]]; then

base_name=$(basename "${wheel}" .whl)

new_name="${base_name}.cu${BUILD_PREV_CUDA_MAJOR}.whl"

mv "${wheel}" "${{ env.CUDA_CORE_ARTIFACTS_DIR }}/cu${BUILD_PREV_CUDA_MAJOR}/${new_name}"

echo "Renamed wheel to: ${new_name}"

fi

done

Once this PR (#1085) is revived/merged, we will be able to build cuda-core in a pure-wheel env and no need to use conda.

leofang · 2025-12-16T05:05:01Z

cuda_core/build_hooks.py

+    # Add local include directory for cuda/core/include
+    local_include_dirs = ["cuda/core"]


Q: Why is this needed? We do not keep any headers under cuda/core or cuda/core/experimental. If without adding this Cython runs into compilation errors, I suspect this has to do with the __init_experimental.pxd file below.

rwgk · 2025-12-16T08:47:05Z

I'm continuing the work under the new PR #1377

I think it was the right decision to start over. Many things got lost in the repeated git merge main steps here:

Summary of Differences: cuda_core_experimental_deprecation_v0 vs cuda_core_experimental_deprecation_v1
========================================================================================================

This summary compares the cuda_core/ directory between the two branches to identify
important changes that might need to be transferred from v0 to v1.

KEY DIFFERENCES IDENTIFIED:
============================

1. **StridedLayout Naming**
   - v0: Uses public `StridedLayout` class
   - v1: Uses private `_StridedLayout` class (with leading underscore)
   - Impact: This is intentional - v1 follows main branch which privatized StridedLayout
   - Status: v1 is correct (matches main branch)

2. **New Memory Resources (from main)**
   - v1 has: `ManagedMemoryResource`, `ManagedMemoryResourceOptions`, 
             `PinnedMemoryResource`, `PinnedMemoryResourceOptions`
   - v0: Missing these
   - Impact: v1 includes newer features from main branch
   - Status: v1 is ahead (has newer features)

3. **StridedMemoryView Exports**
   - v1: Exports `StridedMemoryView` and `args_viewable_as_strided_memory` from `cuda.core`
   - v0: Missing these exports
   - Impact: v1 makes these utilities available at top level
   - Status: v1 is correct

4. **Device.get_all_devices() Method**
   - v1: Has `Device.get_all_devices()` classmethod
   - v0: Missing this method
   - Impact: v1 provides convenience method to get all devices
   - Status: v1 is ahead (has newer feature)

5. **CUDA_CORE_BUILD_MAJOR Type**
   - v1: Uses `int(_get_proper_cuda_bindings_major_version())` (integer comparison)
   - v0: Uses `_get_proper_cuda_bindings_major_version()` (string comparison)
   - v1: Cython code uses `IF CUDA_CORE_BUILD_MAJOR == 12:` (integer)
   - v0: Cython code uses `IF CUDA_CORE_BUILD_MAJOR == "12":` (string)
   - Impact: Type consistency fix
   - Status: v1 is correct (integer comparison is more appropriate)

6. **Include Directory Path**
   - v1: Uses `_include/` (with leading underscore for implementation detail)
   - v0: Uses `include/` (without underscore)
   - Impact: v1 follows the design decision to mark headers as implementation detail
   - Status: v1 is correct (matches user's requirement)

7. **pyproject.toml Package Data**
   - v1: `"cuda.core._include"` 
   - v0: `"cuda.core.include"`
   - Impact: Matches the _include directory rename
   - Status: v1 is correct

8. **build_hooks.py Comments**
   - v1: Updated comment to reference `_include/` instead of `include/`
   - v0: Older comment
   - Impact: Documentation accuracy
   - Status: v1 is correct

STATISTICS:
==========
- ~131 Python/Cython files changed
- Most changes are due to:
  * File moves (experimental/ -> root)
  * StridedLayout -> _StridedLayout renaming
  * Include path updates (include/ -> _include/)
  * New memory resource types added

CONCLUSION:
===========

The v1 branch appears to be MORE UP-TO-DATE than v0, incorporating:
- New memory resource types (ManagedMemoryResource, PinnedMemoryResource)
- Device.get_all_devices() convenience method
- Proper type handling for CUDA_CORE_BUILD_MAJOR
- Correct include directory naming (_include/)
- Proper exports of StridedMemoryView utilities

The main difference is that v0 still uses public `StridedLayout` while v1 uses private
`_StridedLayout`, but this is intentional as v1 follows the main branch which
privatized this class.

RECOMMENDATION:
===============
No important changes need to be transferred from v0 to v1. The v1 branch is
more complete and up-to-date with main branch changes. The v0 branch appears
to be an older incremental migration attempt, while v1 is a fresh, complete
migration that properly incorporates all the latest changes from main.

leofang · 2025-12-17T18:15:12Z

Closing in favor of #1377.

Removed preview folders for the following PRs: - PR #1298 - PR #1397 - PR #1399 - PR #1401

rwgk added 8 commits December 1, 2025 16:24

Update Cython test file to use new import paths

29e31df

[UNTESTED] Update the remaining .pyx test file to use cuda.core.* imports.

Update example files to use new cuda.core.* imports

55a50e6

[UNTESTED] Update all example files to import from cuda.core instead of cuda.core.experimental. Examples should demonstrate the new recommended import paths.

rwgk added 8 commits December 1, 2025 16:48

Fix build_hooks.py to use new cuda.core.* paths

b2f28e9

[UNTESTED] Update build_hooks.py to look for Cython extensions in cuda/core/ instead of cuda/core/experimental/. This fixes the build system to work with the moved files.

Fix _context.pyx to use new import path

45c92f5

[UNTESTED] Update _context.pyx to import from cuda.core._utils.cuda_utils instead of cuda.core.experimental._utils.cuda_utils. This fixes a circular import issue during module initialization.

Fix _memory/_legacy.py to use new import path

b7c0cc0

[UNTESTED] Update _memory/_legacy.py to import from cuda.core._utils.cuda_utils instead of cuda.core.experimental._utils.cuda_utils. This fixes a circular import issue.

Fix _memory/_virtual_memory_resource.py to use new import path

d47b29d

[UNTESTED] Update _memory/_virtual_memory_resource.py to import from cuda.core._utils.cuda_utils instead of cuda.core.experimental._utils.cuda_utils. This fixes a circular import issue.

Fix test helpers to use new cuda.core.* import paths

2e0da66

[UNTESTED] Update cuda_python_test_helpers to import from cuda.core._utils instead of cuda.core.experimental._utils.

Fix test files to use new cuda.core._* import paths

c414b7d

[UNTESTED] Update test files to access internal modules via cuda.core._* instead of cuda.core.experimental._*. This fixes test failures after the namespace migration.

Fix experimental.utils submodule access and test_device_id

d582448

[UNTESTED] Make utils accessible as cuda.core.experimental.utils submodule for backward compatibility. Also fix test_device_id to use cuda.core.system instead of cuda.core.experimental.system.

Fix test_experimental_direct_imports to handle module caching

3cc287c

[UNTESTED] Clear cached experimental module before testing to ensure warnings are emitted. Also improve assertion message.

kkraus14 reviewed Dec 2, 2025

View reviewed changes

rwgk added 3 commits December 1, 2025 21:00

pre-commit fixes (automatic, trivial)

6718a70

Fix import path in test_memory_peer_access.py

506b912

Update import from cuda.core.experimental._utils.cuda_utils to cuda.core._utils.cuda_utils to match the new module structure.

rwgk self-assigned this Dec 2, 2025

rwgk force-pushed the cuda_core_experimental_deprecation_v0 branch from 45fa0bc to dfa0f0e Compare December 2, 2025 06:00

Merge main into cuda_core_experimental_deprecation_v0

64a2b52

Resolved conflicts by updating imports from cuda.core.experimental.* to cuda.core.* to match the branch's deprecation work. Updated all source files and test files to use the new import paths.

leofang reviewed Dec 9, 2025

View reviewed changes

cuda_core/cuda/core/experimental/__init__.py Outdated Show resolved Hide resolved

leofang reviewed Dec 9, 2025

View reviewed changes

rparolin reviewed Dec 11, 2025

View reviewed changes

cuda_core/cuda/core/_memory/_buffer.pyx Show resolved Hide resolved

rparolin reviewed Dec 11, 2025

View reviewed changes

cuda_core/cuda/core/experimental/__init__.py Show resolved Hide resolved

rwgk added 5 commits December 15, 2025 08:31

Sort __getattr__ name list alphabetically and remove one duplicate (_…

4b9796b

…launch_config).

Remove redundant TYPE_CHECKING block from experimental/__init__.py

798ce3d

The TYPE_CHECKING block was unnecessary since all symbols are already imported at runtime below. Type checkers will see the types from the runtime imports, making the TYPE_CHECKING block redundant duplication.

leofang reviewed Dec 16, 2025

View reviewed changes

rwgk mentioned this pull request Dec 16, 2025

cuda-core experimental namespace deprecation (attempt v1) #1377

Merged

leofang removed this from the cuda.core beta 10 milestone Dec 16, 2025

leofang closed this Dec 17, 2025

leofang removed a link to an issue Dec 17, 2025

Deprecate cuda.core.experimental submodule and expose everything to cuda.core main module #1001

Closed

rwgk deleted the cuda_core_experimental_deprecation_v0 branch December 18, 2025 03:39

github-actions bot pushed a commit that referenced this pull request Dec 18, 2025

Clean up PR preview folders for 4 closed/merged PRs

b204d86

Removed preview folders for the following PRs: - PR #1298 - PR #1397 - PR #1399 - PR #1401

	# Rename wheel to include CUDA version suffix
	mkdir -p "${{ env.CUDA_CORE_ARTIFACTS_DIR }}/cu${BUILD_PREV_CUDA_MAJOR}"
	for wheel in ${{ env.CUDA_CORE_ARTIFACTS_DIR }}/*.whl; do
	if [[ -f "${wheel}" ]]; then
	base_name=$(basename "${wheel}" .whl)
	new_name="${base_name}.cu${BUILD_PREV_CUDA_MAJOR}.whl"
	mv "${wheel}" "${{ env.CUDA_CORE_ARTIFACTS_DIR }}/cu${BUILD_PREV_CUDA_MAJOR}/${new_name}"
	echo "Renamed wheel to: ${new_name}"
	fi
	done

		# Add local include directory for cuda/core/include
		local_include_dirs = ["cuda/core"]

cuda-core experimental namespace deprecation (attempt v0) #1298

cuda-core experimental namespace deprecation (attempt v0) #1298

Uh oh!

Conversation

rwgk commented Dec 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

copy-pr-bot bot commented Dec 2, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rwgk commented Dec 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rwgk commented Dec 2, 2025

Uh oh!

rwgk commented Dec 2, 2025

Uh oh!

github-actions bot commented Dec 2, 2025

Preview will be ready when the GitHub Pages deployment is complete.

Uh oh!

Andy-Jost commented Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rwgk commented Dec 9, 2025

Uh oh!

rwgk commented Dec 9, 2025

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rwgk commented Dec 10, 2025

Uh oh!

Uh oh!

Uh oh!

rwgk commented Dec 15, 2025

Uh oh!

rwgk commented Dec 16, 2025

Uh oh!

rwgk commented Dec 16, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rwgk commented Dec 16, 2025

Uh oh!

leofang commented Dec 17, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

rwgk commented Dec 2, 2025 •

edited

Loading

rwgk commented Dec 2, 2025 •

edited

Loading

Andy-Jost commented Dec 3, 2025 •

edited

Loading