[Mlir-commits] [mlir] [mlir][mesh, mpi] More on MeshToMPI (PR #129048)

[Christian Ulmann]

================
@@ -36,10 +42,32 @@ namespace mlir {
 } // namespace mlir
 
 using namespace mlir;
-using namespace mlir::mesh;
+using namespace mesh;
 
 namespace {
-// Create operations converting a linear index to a multi-dimensional index
+/// Convert vec of OpFoldResults (ints) into vector of Values.
+static SmallVector<Value> getMixedAsValues(OpBuilder b, const Location &loc,
+                                           llvm::ArrayRef<int64_t> statics,
+                                           ValueRange dynamics,
+                                           Type type = Type()) {
+  SmallVector<Value> values;
+  auto dyn = dynamics.begin();
+  Type i64 = b.getI64Type();
+  if (!type)
+    type = i64;
+  assert(i64 == type || b.getIndexType() == type);
+  for (auto s : statics) {
+    values.emplace_back(
+        ShapedType::isDynamic(s)
+            ? *(dyn++)
+            : b.create<arith::ConstantOp>(loc, type,
+                                          i64 == type ? b.getI64IntegerAttr(s)
+                                                      : b.getIndexAttr(s)));
----------------
Dinistro wrote:

I suggest to write this with if else, as doubly nested ternary operators are a pain to parse visually.

https://github.com/llvm/llvm-project/pull/129048