Fix rendered docs linkcheck failures

Author: kkraus14

.github/workflows/build-docs.yml

@@ -306,7 +306,7 @@ jobs:
         env:
           PREVIEW_URL: https://nvidia.github.io/cuda-python/pr-preview/pr-${{ env.PR_NUMBER }}
         run: |
-          find "${GITHUB_WORKSPACE}/artifacts/docs" -type f -name '*.html' -printf '%P\n' \
+          find "${GITHUB_WORKSPACE}/artifacts/docs" -type f -name '*.html' ! -path '*/_static/*' -printf '%P\n' \
             | LC_ALL=C sort \
             | sed "s#^#${PREVIEW_URL}/#" > lychee-preview-urls.txt
           if [[ ! -s lychee-preview-urls.txt ]]; then

cuda_bindings/docs/source/conf.py

@@ -37,6 +37,15 @@ def _github_examples_ref():
 GITHUB_EXAMPLES_REF = _github_examples_ref()
 
 
+def _html_baseurl():
+    docs_domain = os.environ.get("CUDA_PYTHON_DOCS_DOMAIN", "https://nvidia.github.io/cuda-python")
+    if int(os.environ.get("BUILD_PREVIEW", 0)):
+        return f"{docs_domain}/pr-preview/pr-{os.environ['PR_NUMBER']}/cuda-bindings/latest/"
+    if int(os.environ.get("BUILD_LATEST", 0)):
+        return f"{docs_domain}/cuda-bindings/latest/"
+    return f"{docs_domain}/cuda-bindings/{release}/"
+
+
 # -- General configuration ---------------------------------------------------
 
 # Add any Sphinx extension module names here, as strings. They can be
@@ -76,7 +85,7 @@ def _github_examples_ref():
 
 # The theme to use for HTML and HTML Help pages.  See the documentation for
 # a list of builtin themes.
-html_baseurl = "docs"
+html_baseurl = _html_baseurl()
 html_theme = "nvidia_sphinx_theme"
 html_theme_options = {
     "switcher": {

cuda_bindings/docs/source/install.rst

@@ -1,4 +1,4 @@
-.. SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+.. SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 .. SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
 Installation
@@ -99,7 +99,7 @@ Source builds require that the provided CUDA headers are of the same major.minor
 
    $ export CUDA_PATH=/usr/local/cuda
 
-See `Environment Variables <environment_variables.rst>`_ for a description of other build-time environment variables.
+See :doc:`Environment Variables <environment_variables>` for a description of other build-time environment variables.
 
 .. note::
 

cuda_bindings/docs/source/module/driver.rst

@@ -406,25 +406,25 @@ Data types used by CUDA driver
     .. autoattribute:: cuda.bindings.driver.CUstreamWaitValue_flags.CU_STREAM_WAIT_VALUE_GEQ
 
 
-        Wait until (int32_t)(*addr - value) >= 0 (or int64_t for 64 bit values). Note this is a cyclic comparison which ignores wraparound. (Default behavior.)
+        Wait until (int32_t)(\*addr - value) >= 0 (or int64_t for 64 bit values). Note this is a cyclic comparison which ignores wraparound. (Default behavior.)
 
 
     .. autoattribute:: cuda.bindings.driver.CUstreamWaitValue_flags.CU_STREAM_WAIT_VALUE_EQ
 
 
-        Wait until *addr == value.
+        Wait until \*addr == value.
 
 
     .. autoattribute:: cuda.bindings.driver.CUstreamWaitValue_flags.CU_STREAM_WAIT_VALUE_AND
 
 
-        Wait until (*addr & value) != 0.
+        Wait until (\*addr & value) != 0.
 
 
     .. autoattribute:: cuda.bindings.driver.CUstreamWaitValue_flags.CU_STREAM_WAIT_VALUE_NOR
 
 
-        Wait until ~(*addr | value) != 0. Support for this operation can be queried with :py:obj:`~.cuDeviceGetAttribute()` and :py:obj:`~.CU_DEVICE_ATTRIBUTE_CAN_USE_STREAM_WAIT_VALUE_NOR`.
+        Wait until ~(\*addr | value) != 0. Support for this operation can be queried with :py:obj:`~.cuDeviceGetAttribute()` and :py:obj:`~.CU_DEVICE_ATTRIBUTE_CAN_USE_STREAM_WAIT_VALUE_NOR`.
 
 
     .. autoattribute:: cuda.bindings.driver.CUstreamWaitValue_flags.CU_STREAM_WAIT_VALUE_FLUSH
@@ -506,19 +506,19 @@ Data types used by CUDA driver
     .. autoattribute:: cuda.bindings.driver.CUstreamAtomicReductionOpType.CU_STREAM_ATOMIC_REDUCTION_OP_OR
 
 
-        Performs an atomic OR: *(address) = *(address) | value
+        Performs an atomic OR: \*(address) = \*(address) | value
 
 
     .. autoattribute:: cuda.bindings.driver.CUstreamAtomicReductionOpType.CU_STREAM_ATOMIC_REDUCTION_OP_AND
 
 
-        Performs an atomic AND: *(address) = *(address) & value
+        Performs an atomic AND: \*(address) = \*(address) & value
 
 
     .. autoattribute:: cuda.bindings.driver.CUstreamAtomicReductionOpType.CU_STREAM_ATOMIC_REDUCTION_OP_ADD
 
 
-        Performs an atomic ADD: *(address) = *(address) + value
+        Performs an atomic ADD: \*(address) = \*(address) + value
 
 .. autoclass:: cuda.bindings.driver.CUstreamAtomicReductionDataType
 

cuda_bindings/docs/source/module/runtime.rst

@@ -1,4 +1,4 @@
-.. SPDX-FileCopyrightText: Copyright (c) 2021-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+.. SPDX-FileCopyrightText: Copyright (c) 2021-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 .. SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
 -------
@@ -299,7 +299,7 @@ Data types used by CUDA Runtime
     .. autoattribute:: cuda.bindings.runtime.cudaError_t.cudaErrorIncompatibleDriverContext
 
 
-        This indicates that the current context is not compatible with this the CUDA Runtime. This can only occur if you are using CUDA Runtime/Driver interoperability and have created an existing Driver context using the driver API. The Driver context may be incompatible either because the Driver context was created using an older version of the API, because the Runtime API call expects a primary driver context and the Driver context is not primary, or because the Driver context has been destroyed. Please see :py:obj:`~.Interactions`with the CUDA Driver API" for more information.
+        This indicates that the current context is not compatible with this the CUDA Runtime. This can only occur if you are using CUDA Runtime/Driver interoperability and have created an existing Driver context using the driver API. The Driver context may be incompatible either because the Driver context was created using an older version of the API, because the Runtime API call expects a primary driver context and the Driver context is not primary, or because the Driver context has been destroyed. Please see `Interactions with the CUDA Driver API`_ for more information.
 
 
     .. autoattribute:: cuda.bindings.runtime.cudaError_t.cudaErrorMissingConfiguration
@@ -6403,11 +6403,11 @@ The types ::CUevent and cudaEvent_t are identical and may be used interchangeabl
 
 
 
-The types ::CUarray and struct ::cudaArray * represent the same data type and may be used interchangeably by casting the two types between each other.
+The types ``CUarray`` and ``struct cudaArray *`` represent the same data type and may be used interchangeably by casting the two types between each other.
 
-In order to use a ::CUarray in a CUDA Runtime API function which takes a struct ::cudaArray *, it is necessary to explicitly cast the ::CUarray to a struct ::cudaArray *.
+In order to use a ``CUarray`` in a CUDA Runtime API function which takes a ``struct cudaArray *``, it is necessary to explicitly cast the ``CUarray`` to a ``struct cudaArray *``.
 
-In order to use a struct ::cudaArray * in a CUDA Driver API function which takes a ::CUarray, it is necessary to explicitly cast the struct ::cudaArray * to a ::CUarray .
+In order to use a ``struct cudaArray *`` in a CUDA Driver API function which takes a ``CUarray``, it is necessary to explicitly cast the ``struct cudaArray *`` to a ``CUarray``.
 
 
 

cuda_core/docs/source/conf.py

@@ -38,6 +38,15 @@ def _github_examples_ref():
 GITHUB_EXAMPLES_REF = _github_examples_ref()
 
 
+def _html_baseurl():
+    docs_domain = os.environ.get("CUDA_PYTHON_DOCS_DOMAIN", "https://nvidia.github.io/cuda-python")
+    if int(os.environ.get("BUILD_PREVIEW", 0)):
+        return f"{docs_domain}/pr-preview/pr-{os.environ['PR_NUMBER']}/cuda-core/latest/"
+    if int(os.environ.get("BUILD_LATEST", 0)):
+        return f"{docs_domain}/cuda-core/latest/"
+    return f"{docs_domain}/cuda-core/{release}/"
+
+
 # -- General configuration ---------------------------------------------------
 
 # Add any Sphinx extension module names here, as strings. They can be
@@ -46,6 +55,7 @@ def _github_examples_ref():
 extensions = [
     "sphinx.ext.autodoc",
     "sphinx.ext.autosummary",
+    "sphinx.ext.extlinks",
     "sphinx.ext.napoleon",
     "sphinx.ext.intersphinx",
     "sphinx.ext.extlinks",
@@ -75,7 +85,7 @@ def _github_examples_ref():
 
 # The theme to use for HTML and HTML Help pages.  See the documentation for
 # a list of builtin themes.
-html_baseurl = "docs"
+html_baseurl = _html_baseurl()
 html_theme = "nvidia_sphinx_theme"
 html_theme_options = {
     "switcher": {

cuda_core/docs/source/examples.rst

@@ -12,48 +12,48 @@ workflow.
 Compilation and kernel launch
 -----------------------------
 
-- :cuda-core-example:`vector_add.py`
+- :cuda-core-example:`vector_add.py <vector_add.py>`
   compiles and launches a simple vector-add kernel with CuPy arrays.
-- :cuda-core-example:`saxpy.py`
+- :cuda-core-example:`saxpy.py <saxpy.py>`
   JIT-compiles a templated SAXPY kernel and launches both float and double
   instantiations.
-- :cuda-core-example:`pytorch_example.py`
+- :cuda-core-example:`pytorch_example.py <pytorch_example.py>`
   launches a CUDA kernel with PyTorch tensors and a wrapped PyTorch stream.
 
 Multi-device and advanced launch configuration
 ----------------------------------------------
 
-- :cuda-core-example:`simple_multi_gpu_example.py`
+- :cuda-core-example:`simple_multi_gpu_example.py <simple_multi_gpu_example.py>`
   compiles and launches kernels across multiple GPUs.
-- :cuda-core-example:`thread_block_cluster.py`
+- :cuda-core-example:`thread_block_cluster.py <thread_block_cluster.py>`
   demonstrates thread block cluster launch configuration on Hopper-class GPUs.
-- :cuda-core-example:`tma_tensor_map.py`
+- :cuda-core-example:`tma_tensor_map.py <tma_tensor_map.py>`
   demonstrates Tensor Memory Accelerator descriptors and TMA-based bulk copies.
 
 Linking and graphs
 ------------------
 
-- :cuda-core-example:`jit_lto_fractal.py`
+- :cuda-core-example:`jit_lto_fractal.py <jit_lto_fractal.py>`
   uses JIT link-time optimization to link user-provided device code into a
   fractal workflow at runtime.
-- :cuda-core-example:`cuda_graphs.py`
+- :cuda-core-example:`cuda_graphs.py <cuda_graphs.py>`
   captures and replays a multi-kernel CUDA graph to reduce launch overhead.
 
 Interoperability and memory access
 ----------------------------------
 
-- :cuda-core-example:`memory_ops.py`
+- :cuda-core-example:`memory_ops.py <memory_ops.py>`
   covers memory resources, pinned memory, device transfers, and DLPack interop.
-- :cuda-core-example:`strided_memory_view_cpu.py`
+- :cuda-core-example:`strided_memory_view_cpu.py <strided_memory_view_cpu.py>`
   uses ``StridedMemoryView`` with JIT-compiled CPU code via ``cffi``.
-- :cuda-core-example:`strided_memory_view_gpu.py`
+- :cuda-core-example:`strided_memory_view_gpu.py <strided_memory_view_gpu.py>`
   uses ``StridedMemoryView`` with JIT-compiled GPU code and foreign GPU buffers.
-- :cuda-core-example:`gl_interop_plasma.py`
+- :cuda-core-example:`gl_interop_plasma.py <gl_interop_plasma.py>`
   renders a CUDA-generated plasma effect through OpenGL interop without CPU
   copies.
 
 System inspection
 -----------------
 
-- :cuda-core-example:`show_device_properties.py`
+- :cuda-core-example:`show_device_properties.py <show_device_properties.py>`
   prints a detailed report of the CUDA devices available on the system.

cuda_core/docs/source/getting-started.rst

@@ -32,7 +32,7 @@ Example: Compiling and Launching a CUDA kernel
 ----------------------------------------------
 
 To get a taste for ``cuda.core``, let's walk through a simple example that compiles and launches a vector addition kernel.
-You can find the complete example in :cuda-core-example:`vector_add.py`
+You can find the complete example in :cuda-core-example:`vector_add.py <vector_add.py>`
 and browse the :doc:`examples page <examples>` for the rest of the shipped
 workflows.
 
@@ -80,7 +80,7 @@ Note the use of the ``name_expressions`` parameter to the :meth:`Program.compile
 Next, we retrieve the compiled kernel from the CUBIN and prepare the arguments and kernel configuration.
 We're using `CuPy <https://cupy.dev/>`_ arrays as inputs for this example, but
 you can use PyTorch tensors too (see
-:cuda-core-example:`pytorch_example.py`
+:cuda-core-example:`pytorch_example.py <pytorch_example.py>`
 and the :doc:`examples page <examples>`).
 
 .. code-block:: python

cuda_core/docs/source/interoperability.rst

@@ -70,11 +70,11 @@ a few iterations to ensure correctness.
 for extracting the metadata (such as pointer address, shape, strides, and
 dtype) from any Python objects supporting either CAI or DLPack and returning a
 :class:`~utils.StridedMemoryView` object. See the
-:cuda-core-example:`strided_memory_view_constructors.py`
+:cuda-core-example:`strided_memory_view_constructors.py <strided_memory_view_constructors.py>`
 example for the explicit constructors, or
-:cuda-core-example:`strided_memory_view_cpu.py`
+:cuda-core-example:`strided_memory_view_cpu.py <strided_memory_view_cpu.py>`
 and
-:cuda-core-example:`strided_memory_view_gpu.py`
+:cuda-core-example:`strided_memory_view_gpu.py <strided_memory_view_gpu.py>`
 for decorator-based workflows. This provides a *concrete implementation* to
 both protocols that is **array-library-agnostic**, so that all Python projects
 can just rely on this without either re-implementing (the consumer-side of)

cuda_core/docs/source/release/0.3.1-notes.rst

@@ -31,8 +31,8 @@ None.
 New examples
 ------------
 
-- Add a `CUDA graph <https://github.com/NVIDIA/cuda-python/blob/|cuda_core_github_ref|/cuda_core/examples/cuda_graphs.py>`_ example.
-- Add a `memory resource <https://github.com/NVIDIA/cuda-python/blob/|cuda_core_github_ref|/cuda_core/examples/memory_ops.py>`_ example.
+- Add a :cuda-core-example:`CUDA graph <cuda_graphs.py>` example.
+- Add a :cuda-core-example:`memory resource <memory_ops.py>` example.
 
 
 Fixes and enhancements