This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [v3 PATCH] LWG 2766, LWG 2749
- From: Jonathan Wakely <jwakely at redhat dot com>
- To: Ville Voutilainen <ville dot voutilainen at gmail dot com>
- Cc: libstdc++ <libstdc++ at gcc dot gnu dot org>, "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>
- Date: Fri, 18 Nov 2016 13:54:22 +0000
- Subject: Re: [v3 PATCH] LWG 2766, LWG 2749
- Authentication-results: sourceware.org; auth=none
- References: <CAFk2RUbErzFd3mGZUiBDf1=X3ruEG03_1EvD4CWHFw-y4ctXWQ@mail.gmail.com>
On 17/11/16 23:38 +0200, Ville Voutilainen wrote:
The first patch does everything but the container adaptor parts (which
are wrong in the p/r) and the tuple part (which is icky). The second
part does the tuple parts, and needs some serious RFC. I know
it's ugly as sin, but it works. I don't think we should try to teach
our tuple to give the right answer for is_constructible etc., because
the last attempt at it broke boost and would break reasonable existing
code in addition to just boost - such things are not Stage 3 material,
imnsho.
I agree, but if we'd just refused to support such undefined behaviour
in stage 1 we wouldn't now be in a position of saying we can't change
it in stage 3.
Oh well, I'll review this ASAP.