On Fri Jul 4 13:52:00 2025 +0000, Rémi Bernon wrote:
I think it is better to use attributes which have the proper semantics already and keep them in the attributes of a declspec/var attributes where all attributes are (even though these attributes are normally restricted to types when used in the IDL), rather than integral values arbitrarily inserted in the expression tree which is probably already abused a bit too much.
Where do you propose to store the attributes in make_exprt()?