220 11784 article Path: news.gmane.org!not-for-mail From: David Krauss Newsgroups: gmane.comp.lang.c++.isocpp.proposals Subject: Re: N4078: Rvalue reference overloads for value() method returns object by value Date: Thu, 10 Jul 2014 14:20:10 +0800 Lines: 187 Approved: news@gmane.org Message-ID: References: <4DA018C3-2327-4F55-B7F8-7988EC7B789A@gmail.com> <1a858a77-310b-4591-aaf5-960c40d9f5fc@isocpp.org> <66AF41C1-CB37-4311-98EA-D47DBC6068DC@gmail.com> <66c38b61-7a95-4ace-b34f-2346bb334fb9@isocpp.org> Reply-To: std-proposals@isocpp.org NNTP-Posting-Host: plane.gmane.org Mime-Version: 1.0 (Mac OS X Mail 7.3 \(1878.6\)) Content-Type: multipart/alternative; boundary="Apple-Mail=_2FA755DB-8D95-40E3-86AF-7CADEF8F0222" X-Trace: ger.gmane.org 1404973229 25747 80.91.229.3 (10 Jul 2014 06:20:29 GMT) X-Complaints-To: usenet@ger.gmane.org NNTP-Posting-Date: Thu, 10 Jul 2014 06:20:29 +0000 (UTC) To: std-proposals@isocpp.org Original-X-From: std-proposals+bncBCW25A7E3QCRBH7B7COQKGQEZ4WGXNA@isocpp.org Thu Jul 10 08:20:21 2014 Return-path: Envelope-to: gclcip-std-proposals@m.gmane.org Original-Received: from mail-pa0-f71.google.com ([209.85.220.71]) by plane.gmane.org with esmtp (Exim 4.69) (envelope-from ) id 1X57iL-0001TU-BG for gclcip-std-proposals@m.gmane.org; Thu, 10 Jul 2014 08:20:17 +0200 Original-Received: by mail-pa0-f71.google.com with SMTP id eu11sf57204096pac.2 for ; Wed, 09 Jul 2014 23:20:16 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:message-id:mime-version:subject:date :references:to:in-reply-to:x-original-sender :x-original-authentication-results:reply-to:precedence:mailing-list :list-id:list-post:list-help:list-archive:list-subscribe :list-unsubscribe:content-type; bh=2jMeU9peKd7K38ITmxcGAidhv3Kp2QWKEMaUJAKyHu4=; b=GTP6h8PcZaYj964OGT1ZQgdT00A4JxLddAyDSddThM7M0ep1f8eRJLWEQJXxrc/2am KE0pMcjRn/lO7GMafUEkeexP3pfluD4qr84zN3L8s4Duc9BSkx6fJEKojwGY/4PWeh4V CmsFwZ898IiDvIwc4EJ75cMv988ZE/mUevthH3wuA7aYP49SyatMN3qoZB42N7r56uX6 nqC5BjWre65tsU1xDZ0OyE7iZqmq4sA8NR3nb2UVjmQBaPWCeO21HPyJ8vVSh6w06hwG ndFyO6iVuoH+LTp980W6qmtKRwlQu+gKaVxHD1N8jTy+X7sKqwet9LfNpFtuFe93ONnv Bh8w== X-Gm-Message-State: ALoCoQnkig5qA+UuXG/AWQYjKy68lVdC5EiBJ+RKAWaYQg3LZ7+FYII+5lT/EC213x0tfPejbS3g X-Received: by 10.70.44.46 with SMTP id b14mr7814161pdm.4.1404973216135; Wed, 09 Jul 2014 23:20:16 -0700 (PDT) X-BeenThere: std-proposals@isocpp.org Original-Received: by 10.50.79.230 with SMTP id m6ls2761igx.39.gmail; Wed, 09 Jul 2014 23:20:15 -0700 (PDT) X-Received: by 10.51.17.97 with SMTP id gd1mr19530104igd.18.1404973215427; Wed, 09 Jul 2014 23:20:15 -0700 (PDT) Original-Received: from mail-ie0-x234.google.com (mail-ie0-x234.google.com [2607:f8b0:4001:c03::234]) by mx.google.com with ESMTPS id b11si74884590icj.83.2014.07.09.23.20.15 for (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Wed, 09 Jul 2014 23:20:15 -0700 (PDT) Received-SPF: pass (google.com: domain of potswa@gmail.com designates 2607:f8b0:4001:c03::234 as permitted sender) client-ip=2607:f8b0:4001:c03::234; Original-Received: by mail-ie0-f180.google.com with SMTP id at20so612238iec.11 for ; Wed, 09 Jul 2014 23:20:15 -0700 (PDT) X-Received: by 10.50.225.98 with SMTP id rj2mr18950785igc.42.1404973215247; Wed, 09 Jul 2014 23:20:15 -0700 (PDT) Original-Received: from [172.20.10.2] ([121.54.54.63]) by mx.google.com with ESMTPSA id n17sm22334738igk.19.2014.07.09.23.20.13 for (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Wed, 09 Jul 2014 23:20:14 -0700 (PDT) In-Reply-To: <66c38b61-7a95-4ace-b34f-2346bb334fb9@isocpp.org> X-Mailer: Apple Mail (2.1878.6) X-Original-Sender: potswa@gmail.com X-Original-Authentication-Results: mx.google.com; spf=pass (google.com: domain of potswa@gmail.com designates 2607:f8b0:4001:c03::234 as permitted sender) smtp.mail=potswa@gmail.com; dkim=pass header.i=@gmail.com; dmarc=pass (p=NONE dis=NONE) header.from=gmail.com Precedence: list Mailing-list: list std-proposals@isocpp.org; contact std-proposals+owners@isocpp.org List-ID: X-Google-Group-Id: 399137483710 List-Post: , List-Help: , List-Archive: List-Subscribe: , List-Unsubscribe: , Xref: news.gmane.org gmane.comp.lang.c++.isocpp.proposals:11784 Archived-At: --Apple-Mail=_2FA755DB-8D95-40E3-86AF-7CADEF8F0222 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=UTF-8 On 2014=E2=80=9307=E2=80=9310, at 2:10 PM, tomaszkam@gmail.com wrote: >=20 > W dniu czwartek, 10 lipca 2014 07:48:29 UTC+2 u=C5=BCytkownik David Kraus= s napisa=C5=82: >=20 > On 2014=E2=80=9307=E2=80=9310, at 1:11 PM, toma...@gmail.com wrote: >=20 >> W dniu czwartek, 10 lipca 2014 01:17:39 UTC+2 u=C5=BCytkownik David Krau= ss napisa=C5=82: >>=20 >> auto && has some strong proponents, and although nothing is a silver bul= let, it will always work except when a function returns an rvalue reference= =E2=80=94 which, as the title of this thread and most of the early message= s indicate, is a problem to be fixed. >=20 >> No, the auto&& does not work with any function that return a any kind of= reference to subobject of it parameter (including this) is invoked on an r= value. In the examples above you my find: >> std::vector f(); >> auto&& s =3D f.front(); >> //This creates a dangling lvalue reference, and front() has not r-value = overloads. >=20 > I think you mean f().front(). As per this thread, or maybe it=E2=80=99s j= ust my extrapolation, the solution is to add an rvalue-ref qualified overlo= ad returning by value. >=20 > As a very general policy in C++ library design, every accessor to a stric= tly owned resource should be lvalue-qualified and/or return by value. This = best practice should be applied across the standard library. >=20 > And dou you expect the following to also work (example presented above): = =20 > auto&& s =3D *f.begin(); Assuming you mean * ( f().begin() ), I would expect that to be currently br= oken, but the fix is to forbid begin() and end() on an xvalue because they = are fundamentally nonsense. It=E2=80=99s a minor issue that some would-be i= ll-formed code is currently accepted. > In addition: "Of course binding a reference to a subobject extends the co= mplete object. What else do you want, destruction of the rest of the object= *except* the subobject??" > Llets assume following scenario: > struct Result > { > std::string array[100]; > std::string& best; //best points to one element of array, making it b= y reference stop at least assigment > };=20 >=20 > //Lets assume that the following class returns array of 100 elements and = hint for the best element > Result foo(); >=20 > //Then you would like the following to work > auto&& cs =3D foo().best; No, lifetime extension doesn=E2=80=99t work like that. It=E2=80=99s never b= een transitive. However even in C++98 this worked: std::string const & cs =3D foo().best; // Lifetime extension since C++98 See 14882:1998 =C2=A712.2/5: > The temporary to which the reference is bound or the temporary that is th= e complete object to a subobject of which the temporary is bound persists f= or the lifetime of the reference except as specified below. It does not mention base subobjects specifically. > Please stop this make "auto&&" movment, becuase it is now going insea wit= h breaking existing code, to make this single statment work. The problem li= es in usign the extension of temporaries by binding in block scope. No existing code is getting broken. The semantics of rvalue references are = understood pretty well now and it=E2=80=99s a completely consistent system. --=20 ---=20 You received this message because you are subscribed to the Google Groups "= ISO C++ Standard - Future Proposals" group. To unsubscribe from this group and stop receiving emails from it, send an e= mail to std-proposals+unsubscribe@isocpp.org. To post to this group, send email to std-proposals@isocpp.org. Visit this group at http://groups.google.com/a/isocpp.org/group/std-proposa= ls/. --Apple-Mail=_2FA755DB-8D95-40E3-86AF-7CADEF8F0222 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=UTF-8
On 2014=E2=80= =9307=E2=80=9310, at 2:10 PM, tomasz= kam@gmail.com wrote:


W dniu czwartek, 10 lipca 2014 07= :48:29 UTC+2 u=C5=BCytkownik David Krauss napisa=C5=82:

= On 2014=E2=80=9307=E2=80=9310, at 1:11 PM, toma...@gmail.com wrote:

W dniu czwartek, 10 lipca 2014 01:17:39 UTC+2 u=C5=BCytkownik = David Krauss napisa=C5=82:

auto && has some strong = proponents, and although nothing is a silver bullet, it will always work ex= cept when a function returns an rvalue reference =E2=80=94 which, as the ti= tle of this thread and most of the early messages indicate, is a problem to= be fixed.

No, the auto&& does not work with any fu= nction that return a any kind of reference to subobject of it parameter (in= cluding this) is invoked on an rvalue. In the examples above you my find:std::vector<std::string> f();
auto&& s =3D f.front();//This creates a dangling lvalue reference, and front() has not r-value ov= erloads.

I think you mean f().front(). As per this thread, or maybe it=E2=80= =99s just my extrapolation, the solution is to add an rvalue-ref qualified = overload returning by value.

As a very general pol= icy in C++ library design, every accessor to a strictly owned resource shou= ld be lvalue-qualified and/or return by value. This best practice should be= applied across the standard library.

And dou you expect the following to also work (example present= ed above):  
      auto&& s = =3D *f.begin();

Assuming yo= u mean * ( f().begin() ), I would expect that= to be currently broken, but the fix is to forbid be= gin() and end() on an xvalue because t= hey are fundamentally nonsense. It=E2=80=99s a minor issue that some would-= be ill-formed code is currently accepted.

In addition: "Of course binding a reference to a su= bobject extends the complete object. What else do you want, destruction of the rest of=20 the object *except* the subobject??"
Llets assume following scena= rio:
struct Result
{
    std::string array[100];    std::string& best; //best points to one element of = array, making it by reference stop at least assigment
};

//Lets = assume that the following class returns array of 100 elements and hint for = the best element
Result foo();

//Then you would like the followin= g to work
auto&& cs =3D foo().best;
=

No, lifetime extension doesn=E2=80=99t work like that. = It=E2=80=99s never been transitive. However even in C++98 this worked:

std::string const & cs =3D = foo().best; // Lifetime extension since C++98

See 14882:1998 =C2=A712.2/5:

The temporary to which the reference is bound or th= e temporary that is the complete object to a subobject of which the te= mporary is bound persists for the lifetime of the reference except as = specified below.

It does not mention base = subobjects specifically.

=
Please stop this make "auto&&" movment, becuase it is now goin= g insea with breaking existing code, to make this single statment work. The= problem lies in usign the extension of temporaries by binding in block sco= pe.

No existing code is gettin= g broken. The semantics of rvalue references are understood pretty well now= and it=E2=80=99s a completely consistent system.

=

--

---
You received this message because you are subscribed to the Google Groups &= quot;ISO C++ Standard - Future Proposals" group.
To unsubscribe from this group and stop receiving emails from it, send an e= mail to std-proposa= ls+unsubscribe@isocpp.org.
To post to this group, send email to std-proposals@isocpp.org.
Visit this group at http://groups.google.com/a/isocpp.org/group/std-proposals/<= /a>.
--Apple-Mail=_2FA755DB-8D95-40E3-86AF-7CADEF8F0222-- .