Date: Tue, 5 Mar 2019 16:14:36 -0500
On Tue, Mar 5, 2019 at 4:03 PM Richard Smith <richard_at_[hidden]> wrote:
> Do we anticipate implementations actually providing a chimera of
> different concepts proposals? Does any implementation already do so?
>
I think GCC is more in danger of being a chimera.
>
> I would prefer to have only the one macro that indicates simply
> whether we have TS concepts or standard concepts, and to not split the
> hair any further unless there's actually a need to do so.
>
Standard concepts currently being from 201811?
>
> On Fri, 1 Mar 2019 at 14:51, Hubert Tong
> <hubert.reinterpretcast_at_[hidden]> wrote:
> >
> > I believe these are meaningful as base values assuming an implementation
> is far enough along to set __cpp_concepts, covering `concept` versus
> `concept bool` and the atomic constraint matching through location +
> parameter mapping.
> >
> > __cpp_concepts: 201507 (TS) or 201707 (P0734).
> > __cpp_requires_expr: 201507 (TS) or 201611 (P0266).
> > __cpp_compound_req: 201507 (TS) or 201707 (P0734) or 201811 (P1084).
> Only 201507 allows placeholder types.
> > __cpp_concept_placeholders: 201507 (TS) or 201811 (P1141).
> >
> > On Fri, Mar 1, 2019 at 4:26 PM John Spicer <jhs_at_[hidden]> wrote:
> >>
> >> [ Dropping core reflector ]
> >>
> >> Do you have a recommendation?
> >>
> >> John.
> >>
> >> On Mar 1, 2019, at 12:03 PM, Hubert Tong <
> hubert.reinterpretcast_at_[hidden]> wrote:
> >>
> >> I'm still hoping for some SG10 guidance on whether we want to query on
> what -> Concept does separately from whether Concept auto is available.
> There is also static_assert(C<T>) in the mix somewhere.
> >>
> >> -- HT
> >>
> >> On Fri, Mar 1, 2019 at 8:45 AM John Spicer <jhs_at_[hidden]> wrote:
> >>>
> >>> That seems appropriate to me (copying the core reflector).
> >>>
> >>> John.
> >>>
> >>> > On Mar 1, 2019, at 7:21 AM, Jonathan Wakely <cxx_at_[hidden]> wrote:
> >>> >
> >>> > The working draft doesn't specify a __cpp_concepts macro. I want to
> know if I need to say "concept bool" or just "concept", so I expected to be
> able to do:
> >>> >
> >>> > #if __cpp_concepts
> >>> > template<typename T>
> >>> > concept
> >>> > #if __cpp_concepts <= 201507
> >>> > bool
> >>> > #endif
> >>> > Foo = ...;
> >>> > #endif
> >>> >
> >>> > Should this be handled by a core issue?
> >>> >
> >>> > _______________________________________________
> >>> > Features mailing list
> >>> > Features_at_[hidden]
> >>> > http://www.open-std.org/mailman/listinfo/features
> >>>
> >>> _______________________________________________
> >>> Features mailing list
> >>> Features_at_[hidden]
> >>> http://www.open-std.org/mailman/listinfo/features
> >>
> >>
> > _______________________________________________
> > Features mailing list
> > Features_at_[hidden]
> > http://www.open-std.org/mailman/listinfo/features
>
> Do we anticipate implementations actually providing a chimera of
> different concepts proposals? Does any implementation already do so?
>
I think GCC is more in danger of being a chimera.
>
> I would prefer to have only the one macro that indicates simply
> whether we have TS concepts or standard concepts, and to not split the
> hair any further unless there's actually a need to do so.
>
Standard concepts currently being from 201811?
>
> On Fri, 1 Mar 2019 at 14:51, Hubert Tong
> <hubert.reinterpretcast_at_[hidden]> wrote:
> >
> > I believe these are meaningful as base values assuming an implementation
> is far enough along to set __cpp_concepts, covering `concept` versus
> `concept bool` and the atomic constraint matching through location +
> parameter mapping.
> >
> > __cpp_concepts: 201507 (TS) or 201707 (P0734).
> > __cpp_requires_expr: 201507 (TS) or 201611 (P0266).
> > __cpp_compound_req: 201507 (TS) or 201707 (P0734) or 201811 (P1084).
> Only 201507 allows placeholder types.
> > __cpp_concept_placeholders: 201507 (TS) or 201811 (P1141).
> >
> > On Fri, Mar 1, 2019 at 4:26 PM John Spicer <jhs_at_[hidden]> wrote:
> >>
> >> [ Dropping core reflector ]
> >>
> >> Do you have a recommendation?
> >>
> >> John.
> >>
> >> On Mar 1, 2019, at 12:03 PM, Hubert Tong <
> hubert.reinterpretcast_at_[hidden]> wrote:
> >>
> >> I'm still hoping for some SG10 guidance on whether we want to query on
> what -> Concept does separately from whether Concept auto is available.
> There is also static_assert(C<T>) in the mix somewhere.
> >>
> >> -- HT
> >>
> >> On Fri, Mar 1, 2019 at 8:45 AM John Spicer <jhs_at_[hidden]> wrote:
> >>>
> >>> That seems appropriate to me (copying the core reflector).
> >>>
> >>> John.
> >>>
> >>> > On Mar 1, 2019, at 7:21 AM, Jonathan Wakely <cxx_at_[hidden]> wrote:
> >>> >
> >>> > The working draft doesn't specify a __cpp_concepts macro. I want to
> know if I need to say "concept bool" or just "concept", so I expected to be
> able to do:
> >>> >
> >>> > #if __cpp_concepts
> >>> > template<typename T>
> >>> > concept
> >>> > #if __cpp_concepts <= 201507
> >>> > bool
> >>> > #endif
> >>> > Foo = ...;
> >>> > #endif
> >>> >
> >>> > Should this be handled by a core issue?
> >>> >
> >>> > _______________________________________________
> >>> > Features mailing list
> >>> > Features_at_[hidden]
> >>> > http://www.open-std.org/mailman/listinfo/features
> >>>
> >>> _______________________________________________
> >>> Features mailing list
> >>> Features_at_[hidden]
> >>> http://www.open-std.org/mailman/listinfo/features
> >>
> >>
> > _______________________________________________
> > Features mailing list
> > Features_at_[hidden]
> > http://www.open-std.org/mailman/listinfo/features
>
Received on 2019-03-05 22:14:53