[patch] Finishing touch for the std::error_code transition

[Chandler Carruth]

On Fri, Jun 13, 2014 at 5:59 PM, Rafael Espíndola <
rafael.espindola at gmail.com> wrote:

> > It means that there are std::errc::... values which we will hand to the
> > generic_category virtual methods which would never get handed to them in
> the
> > standard library because in the standard library these values are always
> > paired with system_category.
>
> Not always. Since std::make_error_code will put them in the
> generic_caterory, any method in the generic category has to be able to
> handle them, no?
>

Ahh, I see.

So generic_category will have to be capable of handling them... And the
whole equivalent virtual dispatch thing is to allow these to compare for
equality with random non-portable things...

NM, I think I see how this ends up working now. We sacrifice portable
equality testing of these enums by forcing them into error_code, but
everything still works. And we're OK without the portable equality testing
because we own all the interfaces and can force them to *all* use this enum.

LGTM! Sorry for the confusion.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140613/18bfa5c3/attachment.html>