cpp: a new approach to stub generation

[ERDI Gergo <cactus cactus rulez org>]

so I'm fighting my way throught the horrible mess that is the cpp idl
compiler, and fixing some reinterpret_cast<>'s with sometimes equally
broken solutions, and I've realized that much of the custom code generator
code from the various IDLType subclasses could be removed with adding
three clever new virtual methods to IDLType:

	virtual void stub_c_to_cpp (ostream      &ostr,
				    Indent       &indent,
				    string const &c_obj_name,
				    string const &cpp_obj_name) const = 0;
	virtual void stub_c_to_cpp_ref (ostream      &ostr,
					Indent       &indent,
					string const &c_obj_name,
					string const &cpp_obj_name) const = 0;
	virtual void stub_cpp_to_c (ostream      &ostr,
				    Indent       &indent,
				    string const &cpp_obj_name,
				    string const &c_obj_name) const = 0;

Most of the IDLType subclasses would only need to override these, if the
IDLType methods are written to use these.
For example, lots of subclasses implement writeUnionAccessors, while the
following implementation in IDLType would be enough for all of them:

void
IDLType::writeUnionAccessors (ostream          &ostr,
			      Indent           &indent,
			      string const     &id,
			      string const     &discriminatorVal,
			      IDLTypedef const *activeTypedef = 0) const
{
	string typespec, dcl;
	getCPPStubDeclarator (IDL_PARAM_IN, "", typespec, dcl,
activeTypedef);
	string c_member = "m_target._u." + id;

	ostr << indent << typespec << dcl << " " << id << "() const {" <<
endl;
	stub_c_to_cpp_ref (ostr, ++indent, c_member, "_cpp_retval");
	ostr << indent << "return _cpp_retval;" << endl;
	ostr << --indent << "}" << endl << endl;
}

Should I go on and do this? Comments welcome,

	Gergo

-- 
   .--= ULLA! =---------------------.   `We are not here to give users what
   \     http://cactus.rulez.org     \   they want'  -- RMS, at GUADEC 2001
    `---= cactus@cactus.rulez.org =---'
Ki volt Casper, mielôtt meghalt?