Is xdemangle.hpp redundant ?

[anutosh491]

I see that xdemangle.hpp provides the demangle function.

xeus-cpp/src/xdemangle.hpp

Lines 41 to 42 in e9df705

	const char* demangle(const char* name) noexcept;
	const char* demangle(const std::string& name) noexcept;

This should technically be used in xinspect in the implementation of find_type (as done for xeus-cling)

But we are currently using find_type_slow (the function name suggests this might not be the best approach ?!)

xeus-cpp/src/xinspect.hpp

Lines 82 to 99 in e9df705

	std::string find_type_slow(const std::string& expression) {
	static unsigned long long var_count = 0;
	if (auto *type = Cpp::GetType(expression))
	return Cpp::GetQualifiedName(type);
	// Here we might need to deal with integral types such as 3.14.
	std::string id = "__Xeus_GetType_" + std::to_string(var_count++);
	std::string using_clause = "using " + id + " = __typeof__(" + expression + ");\n";
	if (!Cpp::Declare(using_clause.c_str(), /*silent=*/false)) {
	Cpp::TCppScope_t lookup = Cpp::GetNamed(id, nullptr);
	Cpp::TCppType_t lookup_ty = Cpp::GetTypeFromScope(lookup);
	return Cpp::GetQualifiedCompleteName(Cpp::GetCanonicalType(lookup_ty));
	}
	return "";
	}

But if we are not trying to implement a "find_type_fast" and are comfortable with what we have, let's get rid of xdemangle.hpp ?

[vgvassilev]

Yes, it can go away.