[Mlir-commits] [mlir] [mlir] Add unit test for RankedTensorType wrapper class example. (PR #99789)

[River Riddle]

================
@@ -226,4 +227,63 @@ TEST(ShapedTypeTest, RankedTensorTypeBuilder) {
   }
 }
 
+/// Simple wrapper class to enable "isa querying" and simple accessing of
+/// encoding.
+class TensorWithString : public RankedTensorType {
+public:
+  using RankedTensorType::RankedTensorType;
+
+  static TensorWithString get(ArrayRef<int64_t> shape, Type elementType,
+                              StringRef name) {
+    return mlir::cast<TensorWithString>(RankedTensorType::get(
+        shape, elementType, StringAttr::get(elementType.getContext(), name)));
+  }
+
+  StringRef getName() const {
+    if (Attribute enc = getEncoding())
+      return mlir::cast<StringAttr>(enc).getValue();
----------------
River707 wrote:

If we're going to be adding these encoding wrappers, should we consider a templatized get encoding that does the cast/dyn_cast to the wrapped type?

https://github.com/llvm/llvm-project/pull/99789