new character types [was: [Patch] libstdc++/32907]

["Benjamin Kosnik" <bkoz at redhat dot com>]

> >I would like to see a serious effort on behalf of the new character
> >types, as soon as possible.
> >
> Yes, I would be also interested, in principle. Generally, something that
> almost unconsciously, was blocking me on that (library) project, is that
> I was under the impression that some front-end changes are necessary, to
> support char32_t and char16_t as "first class" types. But I should
> really study the various documents more carefully. Are you clear on
> those basic issues?

I have not studied the implementation space at all. The char32_t and
char16_t issues seem, at first blush, to be somewhat simpler than the
corresponding wchar_t implementation, since the size of the type is
exactly specified. The devil tends to be in the details with this
stuff...

I think that the bug reports from so.6's std::string make it clear
that non char/wchar_t instantiations have some issues. I would suggest
that the char32_t and char16_t instantiations of std::string use versa
string from the start.... thoughts? I'm not quite sure how to mix
stuff like this though.

You are correct in that adding new fundamental types will need to be
done in the FE, and that these new types impact overload resolution,
etc.

best,
benjamin