address comments

Author: lmondada

runtime/common/CompiledKernel.cpp

@@ -34,7 +34,7 @@ cudaq::CompiledKernel::execute(const std::vector<void *> &rawArgs) const {
     return reinterpret_cast<KernelThunkResultType (*)(void *, bool)>(funcPtr)(
         buff, /*client_server=*/false);
   }
-  if (jitRepr && jitRepr->argsCreator) {
+  if (!isFullySpecialized()) {
     void *buff = nullptr;
     jitRepr->argsCreator(static_cast<const void *>(rawArgs.data()), &buff);
     reinterpret_cast<KernelThunkResultType (*)(void *, bool)>(funcPtr)(
@@ -48,7 +48,7 @@ cudaq::CompiledKernel::execute(const std::vector<void *> &rawArgs) const {
 }
 
 cudaq::KernelThunkResultType cudaq::CompiledKernel::execute() const {
-  if (jitRepr && jitRepr->argsCreator)
+  if (!isFullySpecialized())
     throw std::runtime_error(
         "Kernel has unspecialized parameters; call execute(rawArgs) instead.");
   if (!resultInfo.hasResult()) {

runtime/common/CompiledKernel.h

@@ -133,7 +133,9 @@ class CompiledKernel {
   const MlirRepr &getMlir() const;
 
   std::string name;
-  ResultInfo resultInfo;
+  ResultInfo resultInfo; // TODO: we might want to store the entire kernel
+                         // signature here. Though I'm not sure what MLIR
+                         // agnostic information is worth storing.
   std::optional<JitRepr> jitRepr;
   std::optional<MlirRepr> mlirRepr;
 };