This avoids parsing the wrong file to display the error.

Py_CompileString(..., ..., Py_file_input) parses the file to display the traceback error. I don't know why if it cannot get the file, then it uses the inline script import. Thus, it is displaying the wrong traceback, because it is parsing the wrong file.
To avoid that, I set the file name to <In [1]>, so when executing codes in Jupyter, it is displaying the error correctly.

Author: cindytsai

src/libyt_kernel.cpp

@@ -68,7 +68,7 @@ nl::json LibytKernel::execute_request_impl(int execution_counter, const std::str
                                            bool store_history, nl::json user_expressions, bool allow_stdin) {
     SET_TIMER(__PRETTY_FUNCTION__);
 
-    std::string cell_name = std::string("In [") + std::to_string(execution_counter) + std::string("]");
+    std::string cell_name = std::string("<In [") + std::to_string(execution_counter) + std::string("]>");
 
     // Find if '%' is the first non-space character, if so, redirect jobs to define command
     std::size_t found = code.find_first_not_of("\t\n\v\f\r ");