Merge pull request #361 from bdice/branch-0.43-merge-0.42

Forward-merge branch-0.42 into branch-0.43

README.md

@@ -15,7 +15,7 @@ $ conda env create -n ucxx -f conda/environments/all_cuda-118_arch-x86_64.yaml
 Or for CUDA 12.x:
 
 ```
-$ conda env create -n ucxx -f conda/environments/all_cuda-125_arch-x86_64.yaml
+$ conda env create -n ucxx -f conda/environments/all_cuda-128_arch-x86_64.yaml
 ```
 
 And then activate the newly created environment:

conda/environments/all_cuda-118_arch-x86_64.yaml

@@ -25,7 +25,7 @@ dependencies:
 - librmm==25.4.*,>=0.0.0a0
 - libtool
 - ninja
-- numba>=0.59.1.1,<0.61.0a0
+- numba>=0.59.1,<0.61.0a0
 - numpy>=1.23,<3.0a0
 - pip
 - pkg-config

conda/environments/all_cuda-125_arch-x86_64.yaml → conda/environments/all_cuda-128_arch-x86_64.yaml

@@ -13,7 +13,7 @@ dependencies:
 - cloudpickle
 - cmake>=3.26.4,!=3.30.0
 - cuda-cudart-dev
-- cuda-version=12.5
+- cuda-version=12.8
 - cudf==25.4.*,>=0.0.0a0
 - cupy>=12.0.0
 - cxx-compiler
@@ -25,7 +25,7 @@ dependencies:
 - librmm==25.4.*,>=0.0.0a0
 - libtool
 - ninja
-- numba>=0.59.1.1,<0.61.0a0
+- numba>=0.59.1,<0.61.0a0
 - numpy>=1.23,<3.0a0
 - pip
 - pkg-config
@@ -42,4 +42,4 @@ dependencies:
 - setuptools>=64.0.0
 - spdlog>=1.14.1,<1.15
 - ucx>=1.15.0,<1.19
-name: all_cuda-125_arch-x86_64
+name: all_cuda-128_arch-x86_64

conda/recipes/ucxx/meta.yaml

@@ -231,7 +231,7 @@ outputs:
         - pynvml >=12.0.0,<13.0.0a0
       run_constrained:
         - cupy >=9.5.0
-        - numba >=0.59.1.0,<0.61.0a0
+        - numba >=0.59.1,<0.61.0a0
     test:
       commands:
         - test -f $PREFIX/lib/libucxx_python.so

cpp/include/ucxx/endpoint.h

@@ -55,9 +55,9 @@ class Endpoint : public Component {
   bool _endpointErrorHandling{true};  ///< Whether the endpoint enables error handling
   std::unique_ptr<InflightRequests> _inflightRequests{
     std::make_unique<InflightRequests>()};  ///< The inflight requests
-  std::mutex _mutex{std::mutex()};  ///< Mutex used during close to prevent race conditions between
-                                    ///< application thread and `ucxx::Endpoint::setCloseCallback()`
-                                    ///< that may run asynchronously on another thread.
+  std::mutex _mutex{};  ///< Mutex used during close to prevent race conditions between
+                        ///< application thread and `ucxx::Endpoint::setCloseCallback()`
+                        ///< that may run asynchronously on another thread.
   ucs_status_t _status{UCS_INPROGRESS};  ///< Endpoint status
   std::atomic<bool> _closing{false};     ///< Prevent calling close multiple concurrent times.
   EndpointCloseCallbackUserFunction _closeCallback{nullptr};  ///< Close callback to call

dependencies.yaml

@@ -3,7 +3,7 @@ files:
   all:
     output: conda
     matrix:
-      cuda: ["11.8", "12.5"]
+      cuda: ["11.8", "12.8"]
       arch: [x86_64]
     includes:
       - build_cpp
@@ -225,6 +225,10 @@ dependencies:
               cuda: "12.5"
             packages:
               - cuda-version=12.5
+          - matrix:
+              cuda: "12.8"
+            packages:
+              - cuda-version=12.8
   cuda:
     specific:
       - output_types: conda
@@ -288,7 +292,7 @@ dependencies:
     common:
       - output_types: [conda, requirements, pyproject]
         packages:
-          - &numba numba>=0.59.1.1,<0.61.0a0
+          - &numba numba>=0.59.1,<0.61.0a0
           - rapids-dask-dependency==25.4.*,>=0.0.0a0
   test_cpp:
     common:

python/distributed-ucxx/pyproject.toml

@@ -16,7 +16,7 @@ authors = [
 license = { text = "BSD-3-Clause" }
 requires-python = ">=3.10"
 dependencies = [
-    "numba>=0.59.1.1,<0.61.0a0",
+    "numba>=0.59.1,<0.61.0a0",
     "rapids-dask-dependency==25.4.*,>=0.0.0a0",
     "ucxx==0.43.*,>=0.0.0a0",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.

python/ucxx/pyproject.toml

@@ -41,7 +41,7 @@ test = [
     "cloudpickle",
     "cudf==25.4.*,>=0.0.0a0",
     "cupy-cuda11x>=12.0.0",
-    "numba>=0.59.1.1,<0.61.0a0",
+    "numba>=0.59.1,<0.61.0a0",
     "pytest-asyncio",
     "pytest-rerunfailures",
     "pytest==7.*",