[Mlir-commits] [mlir] [mlir][sparse] Replace `getSparseTensorType` with `tryGetSparseTensorType` (PR #109435)

[Aart Bik]

================
@@ -826,3 +826,19 @@ func.func @sparse_new_coo_permute_no(%arg0: !llvm.ptr) -> tensor<?x?xf32, #CooPN
   %0 = sparse_tensor.new %arg0 : !llvm.ptr to tensor<?x?xf32, #CooPNo>
   return %0 : tensor<?x?xf32, #CooPNo>
 }
+
+// CHECK-LABEL: func.func @test_tensor_dim_unranked
----------------
aartbik wrote:

codegen.mlir is a test that very specifically tests the codegen path

the added tests are really regression tests that test crash-before, no-crash after behavior
as such, can you please put them in a new file and document that

e.g. no_lowering.mlir (or better name)
that way you can also reduce the flags in the runner

https://github.com/llvm/llvm-project/pull/109435