fix compiler errors for modern linux+clang+python+cuda setup

.gitignore

@@ -12,3 +12,4 @@ CMakeCache.txt
 doc
 .idea/
 apps/tensor_times_vector/tensor_times_vector
+*.lock

CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 3.4.0 FATAL_ERROR)
+cmake_minimum_required(VERSION 3.18 FATAL_ERROR)
 if(POLICY CMP0048)
   cmake_policy(SET CMP0048 NEW)
 endif()
@@ -7,7 +7,7 @@ if(POLICY CMP0054)
 endif()
 project(taco
   VERSION 0.1
-  LANGUAGES C CXX
+  LANGUAGES C CXX CUDA
 )
 option(CUDA "Build for NVIDIA GPU (CUDA must be preinstalled)" OFF)
 option(PYTHON "Build TACO for python environment" OFF)
@@ -18,7 +18,7 @@ set(TACO_FEATURE_OPENMP 0)
 set(TACO_FEATURE_PYTHON 0)
 if(CUDA)
   message("-- Searching for CUDA Installation")
-  find_package(CUDA REQUIRED)
+  find_package(CUDAToolkit)
   add_definitions(-DCUDA_BUILT)
   set(TACO_FEATURE_CUDA 1)
 endif(CUDA)
@@ -126,13 +126,20 @@ if(GIT_FOUND AND EXISTS "${TACO_PROJECT_DIR}/.git")
   # Update submodules as needed
   option(GIT_SUBMODULE "Check submodules during build" ON)
   if(GIT_SUBMODULE)
-    message(STATUS "Submodule update")
+    message(STATUS "Updating submodules with shared lock file: ${CMAKE_SOURCE_DIR}/submodules.lock")
+    file(LOCK "${CMAKE_SOURCE_DIR}/submodules.lock" TIMEOUT 120)
     execute_process(COMMAND ${GIT_EXECUTABLE} submodule update --init --recursive
             WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
             RESULT_VARIABLE GIT_SUBMOD_RESULT)
     if(NOT GIT_SUBMOD_RESULT EQUAL "0")
       message(FATAL_ERROR "git submodule update --init failed with ${GIT_SUBMOD_RESULT}, please checkout submodules")
     endif()
+    execute_process(COMMAND ${GIT_EXECUTABLE} checkout stable
+            WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/python_bindings/pybind11
+            RESULT_VARIABLE GIT_PYBIND11_CHECKOUT_RESULT)
+    if(NOT GIT_PYBIND11_CHECKOUT_RESULT EQUAL "0")
+      message(WARNING "git checkout stable failed with ${GIT_SUBMOD_RESULT}, build might be unstable")
+    endif()
   endif()
   # get git revision
   execute_process(

README.md

@@ -29,7 +29,7 @@ TL;DR build taco using CMake. Run `make test`.
 # Build and test
 ![Build and Test](https://github.com/RSenApps/taco/workflows/Build%20and%20Test/badge.svg?branch=master)
 
-Build taco using CMake 3.4.0 or greater:
+Build taco using CMake 3.18 or greater:
 
     cd <taco-directory>
     mkdir build

apps/tensor_times_vector/CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 2.8.12)
+cmake_minimum_required(VERSION 3.10)
 if(POLICY CMP0048)
   cmake_policy(SET CMP0048 NEW)
 endif()

src/CMakeLists.txt

@@ -1,3 +1,5 @@
+cmake_minimum_required(VERSION 3.18 FATAL_ERROR)
+
 if (TACO_SHARED_LIBRARY)
   set(TACO_LIBRARY_TYPE SHARED)
   message("-- Shared library")
@@ -21,8 +23,7 @@ add_definitions(${TACO_DEFINITIONS})
 include_directories(${TACO_SRC_DIR})
 add_library(taco ${TACO_LIBRARY_TYPE} ${TACO_HEADERS} ${TACO_SOURCES})
 if (CUDA)
-  include_directories(${CUDA_INCLUDE_DIRS})
-  target_link_libraries(taco PUBLIC ${CUDA_LIBRARIES})
+  target_link_libraries(taco PUBLIC CUDA::cudart)
 endif (CUDA)
 install(TARGETS taco DESTINATION lib)
 

src/lower/lowerer_impl_imperative.cpp

@@ -497,7 +497,6 @@ Stmt LowererImplImperative::lowerAssignment(Assignment assignment)
     if (isAssembledByUngroupedInsertion(result)) {
       std::vector<Expr> coords;
       Expr prevPos = 0;
-      size_t i = 0;
       const auto resultIterators = getIterators(assignment.getLhs());
       for (const auto& it : resultIterators) {
         // TODO: Should only assemble levels that can be assembled together
@@ -517,7 +516,6 @@ Stmt LowererImplImperative::lowerAssignment(Assignment assignment)
         }
 
         prevPos = pos;
-        ++i;
       }
     }
 

test/tests-scheduling.cpp

@@ -567,8 +567,8 @@ TEST(scheduling, multilevel_tiling) {
 
   vector<IndexVar> reordering = {iX1, iX2, iY1, iY2};
   sort(reordering.begin(), reordering.end());
-  int countCorrect = 0;
-  int countIncorrect = 0;
+  //int countCorrect = 0;
+  //int countIncorrect = 0;
   do {
     // TODO: Precondition (can be broken) bottom most loop must remain unchanged if sparse
     bool valid_reordering = reordering[3] == iY2;
@@ -583,17 +583,15 @@ TEST(scheduling, multilevel_tiling) {
     C.compute();
     if (!equals(C, expected)) {
       //cout << util::join(reordering) << endl;
-      countIncorrect++;
+      //countIncorrect++;
 
       std::shared_ptr<ir::CodeGen> codegen = ir::CodeGen::init_default(cout, ir::CodeGen::ImplementationGen);
       ir::Stmt compute = lower(reordered, "compute",  false, true);
       codegen->compile(compute, true);
       ASSERT_TENSOR_EQ(expected, C);
       exit(1);
     }
-    else {
-      countCorrect++;
-    }
+    //else{countCorrect++;}
   } while (next_permutation(reordering.begin(), reordering.end()));
 }
 

test/tests-tensor_types.cpp

@@ -45,7 +45,7 @@ TYPED_TEST_P(VectorTensorTest, types) {
   ASSERT_EQ(t, a.getComponentType());
   ASSERT_EQ(1, a.getOrder());
   ASSERT_EQ(5, a.getDimension(0));
-  map<vector<int>,TypeParam> vals = {{{0}, 1.0}, {{2}, 2.0}};
+  map<vector<int>,TypeParam> vals = {{{0}, static_cast<TypeParam>(1)}, {{2}, static_cast<TypeParam>(2)}};
   for (auto& val : vals) {
     a.insert(val.first, val.second);
   }