On Mon, Jul 31, 2023 at 4:48 PM Thiago Macieira via Std-Proposals
<std-proposals@lists.isocpp.org> wrote:
>
> What's more, changing this is a full-on language ABI break because it adds a
> new field that isn't there in the type_info. That means the impact of *adding*
> this now is extremely high, equal to replacing std:: with std2.
Ok then here's a compromise to prevent ABI breakage: If you don't put
it inside the 'type_info' object, then at least provide a way of
retrieving it from any 'type_info' object, as follows:
In practice, type_info is not very useful for getting information about types because there is an infinite universe of information you might want. It's useful for comparing two types with each other, so that you can store the information you actually want in a std::map or std::unordered_map and then look it up using the type_info.
(Yes, you can get the mangled name from type_info::name, but there's usually not much you can do with it other than eventually printing it out for a human to read.)
For example, std::any needs to know not just the size of the object that it stores, but also how to copy-construct and how to destroy that object. Will you also propose adding methods `void copy_construct(void* dest, const void* src)` and `void destroy(void* obj)` to the std::type_info class? And if you propose adding those, then where do you draw the line? Will we end up with a dozen compiler switches (which the implementors will have to maintain) so that every user can get only the information they want in their type_info objects?
And you can implement std::any without that kind of support from the language. I'm guessing there's a way to do what you want to do as well, without type_info needing to store the size.
namespace std {
size_t typeinfo_sizeof (type_info const &);
size_t typeinfo_alignof(type_info const &);
};
int main(void)
{
std::type_info const &ti = typeid( SomeClass );
cout << "Alignment of SomeClass: " << typeinfo_alignof(ti) << endl;
}
If you're worried about the overhead then compilers can have a switch
"-fno-extended-typeinfo".
--
Std-Proposals mailing list
Std-Proposals@lists.isocpp.org
https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
--