Am 15. April 2021 14:41:34 MESZ schrieb Aaron Ballman <aaron_at_[hidden]>:
>On Thu, Apr 15, 2021 at 8:02 AM Jens Gustedt <jens.gustedt_at_[hidden]>
>wrote:
>>
>> Aaron,
>>
>> on Thu, 15 Apr 2021 07:08:34 -0400 you (Aaron Ballman
>> <aaron_at_[hidden]>) wrote:
>>
>> > On Thu, Apr 15, 2021 at 6:50 AM Jens Gustedt via Liaison
>> > <liaison_at_[hidden]> wrote:
>> > >
>> > > Ville,
>> > >
>> > > on Thu, 15 Apr 2021 13:22:05 +0300 you (Ville Voutilainen
>> > > <ville.voutilainen_at_[hidden]>) wrote:
>> > ...
>>
>> > > no, these operators are not new, they would just be different
>> > > spellings
>> > >
>> > > (and I described how I see that this fits into the C
>specification
>> > > in several papers, you could have a look at
>> > > http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2644.pdf, for
>> > > example)
>> > > > and add them as alternative spellings of some existing semantic
>> > > > descriptions. Overall, I think there's not much specification
>> > > > difficulty, at least not for things that are just alternate
>> > > > spellings.
>> > >
>> > > That's the kind of information that I was after, thank you.
>> > >
>> > > > I don't know whether you envision adding new operators,
>> > >
>> > > no
>> >
>> > Then these aren't "just different spellings" as far as I can tell.
>For
>> > example, this is valid C++ code that uses a different spelling:
>> >
>> > struct S {
>> > int operator and(int i); // [lex.digraph]
>> > int operator <::>(int i) const; // [lex.digraph]
>> > };
>>
>> yes
>>
>> > Why are these hypothetical punctuators different?
>>
>> Sorry, maybe I don't get what you are asking.
>
>Perhaps I misunderstood, but I thought Ville was asking whether you'd
>be able to write:
>
>int operator∧(int i);
>
>and I thought you were answering that you did not intend for that to
>be a valid operator overload.
>
>> If you'd have `¬a ∧ b` that would just be another spelling of `!a &&
>b`.
>>
>> If you'd have `⟦vendor∷pride⟧` that would effectively be two new
>> tokens `⟦` and `⟧` that the languages do not yet have, compared to
>the
>> one token `∷` that would be another spelling of `::`. We would have
>to
>> have a syntax rule to make this equivalent to `[[vendor::pride]]`.
>>
>> But these tokens would not be operator tokens with new semantic or
>so,
>> just new syntactic notation of some part of a declarator.
>
>My concern is addressed if operator ∧(), operator &&(), and operator
>and() are all identical declarations (except for spelling). e.g., I
>should be able to write:
>
>struct S {
> int operator∧(int) const;
>};
>
>S s;
>s && 12; // ok, calls operator∧()
>
>~Aaron
>
>
>>
>> Jens
>>
>> --
>> :: INRIA Nancy Grand Est ::: Camus ::::::: ICube/ICPS :::
>> :: ::::::::::::::: office Strasbourg : +33 368854536 ::
>> :: :::::::::::::::::::::: gsm France : +33 651400183 ::
>> :: ::::::::::::::: gsm international : +49 15737185122 ::
>> :: http://icube-icps.unistra.fr/index.php/Jens_Gustedt ::

Yes, that's the main idea. For punctuators that correspond to operators this would just be a different spelling.

Jens

-- 
Jens Gustedt - INRIA & ICube, Strasbourg, France