220 5198 <8e156cc1-8136-48e6-96f2-8dca72fee7d4@isocpp.org> article Path: news.gmane.org!not-for-mail From: Xeo Newsgroups: gmane.comp.lang.c++.isocpp.proposals Subject: Re: Teachability problems with std::move Date: Mon, 24 Jun 2013 13:08:29 -0700 (PDT) Lines: 219 Approved: news@gmane.org Message-ID: <8e156cc1-8136-48e6-96f2-8dca72fee7d4@isocpp.org> References: Reply-To: std-proposals@isocpp.org NNTP-Posting-Host: plane.gmane.org Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_296_23524625.1372104509838" X-Trace: ger.gmane.org 1372105131 20065 80.91.229.3 (24 Jun 2013 20:18:51 GMT) X-Complaints-To: usenet@ger.gmane.org NNTP-Posting-Date: Mon, 24 Jun 2013 20:18:51 +0000 (UTC) To: std-proposals@isocpp.org Original-X-From: std-proposals+bncBCCNBYNYXMJRBPWOUKHAKGQESBOETDA@isocpp.org Mon Jun 24 22:18:52 2013 Return-path: Envelope-to: gclcip-std-proposals@m.gmane.org Original-Received: from mail-gh0-f200.google.com ([209.85.160.200]) by plane.gmane.org with esmtp (Exim 4.69) (envelope-from ) id 1UrDDv-0008Mw-Q0 for gclcip-std-proposals@m.gmane.org; Mon, 24 Jun 2013 22:18:52 +0200 Original-Received: by mail-gh0-f200.google.com with SMTP id 10sf11942087ghy.11 for ; Mon, 24 Jun 2013 13:18:50 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=x-beenthere:date:from:to:message-id:in-reply-to:references:subject :mime-version:x-original-sender:reply-to:precedence:mailing-list :list-id:x-google-group-id:list-post:list-help:list-archive :list-subscribe:list-unsubscribe:content-type; bh=jpUXSIRkmzy4sdcmxd3zpVyEaYF7fwwkzqAUfJ8/fho=; b=aDDdz2bmbguaHWh9mSlod5xjpaWgt1Xjv3rXF5cVzTAq6acDazuCEDLnGEf2o5Q4cu vigqgwJlhuwyB584hXs2Cfuw7hNEA133CPM/Y0jMBuXjUKJ9cbCZ3ywVEzI3L8wPzTDJ RhoOH4NCrL6toGclktxxqwBSswUts2mxaxQksGKQA6r1JQYYHLKbtKoZZxkQWfPGcnAH jFfPjCY82nXP4ZTH80au8wnImmyaMFD92af8e3z5WIFUOHD8TcEUJnkl6rY+qaEPmdIU GMO3HZmJFArsxHoVxhZZ26F8OTCoxaupxJmnYf9b3AMLjSgPv/XIEtQpHYydIph12jAi oS7Q== X-Received: by 10.224.205.138 with SMTP id fq10mr27171435qab.1.1372104511314; Mon, 24 Jun 2013 13:08:31 -0700 (PDT) X-BeenThere: std-proposals@isocpp.org Original-Received: by 10.49.61.166 with SMTP id q6ls2129534qer.70.gmail; Mon, 24 Jun 2013 13:08:30 -0700 (PDT) X-Received: by 10.49.62.3 with SMTP id u3mr564963qer.26.1372104510279; Mon, 24 Jun 2013 13:08:30 -0700 (PDT) In-Reply-To: X-Original-Sender: hivemaster@hotmail.de 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:5198 Archived-At: ------=_Part_296_23524625.1372104509838 Content-Type: text/plain; charset=ISO-8859-1 I remember sending you and Mike a reminder about that some months ago, and was wondering, did you actually talk about this at the Bristol meeting? On Monday, June 24, 2013 9:14:17 PM UTC+2, Richard Smith wrote: > > On Mon, Jun 24, 2013 at 11:42 AM, Geoffrey Romer > > wrote: > > It's surprisingly hard to come up with a good, teachable set of rules > for > > when to use std::move. For example, it would be nice to be able to say > "you > > can leave out std::move when returning a named local variable", but this > is > > not always the case. Consider: > > > > std::unique_ptr FooFactory() { > > std::unique_ptr result(new Foo); > > // ... > > return std::move(result); > > } > > > > The "std::move" is mandatory here, because an lvalue can only be > implicitly > > treated as an rvalue during copy operations, and in "return result;", > > 'result' is not copied, but used as the input for an implicit > conversion. It > > seems like this should be fixable, perhaps by expanding the > implicit-rvalue > > rules, but I'm not sure what all the consequences would be (at a > minimum, > > this could change the behavior of existing code that has separate rvalue > and > > lvalue overloads for an implicit conversion operation). > > Yes, for NRVO, I think the "same cv-unqualified type as the function > return type" restriction only needs to apply to copy-elision and not > to implicit move. That is, I think we should treat 'result' as an > xvalue here: > > T f() { > U result; > return result; > } > > ... independent of the types of T and U, but we should only be > permitted to perform copy-elision of they're the same cv-unqualified > type. Most generalizations of the implicit-rvalue rules are fraught > with peril, but this one seems relatively safe. > > > Conversely, it would be nice to be able to say "When in doubt, just use > > std::move if you want move semantics, because it never hurts", but > std::move > > can in fact be a pessimization: > > > > std::array MakeHugeArray() { > > std::array result; > > // ... > > return std::move(result); > > } > > I suggest you talk to your compiler vendor (Hi!) and ask for a warning > for this case. > > > As written, the return statement incurs 10,000 invocations of Foo's move > > constructor (or, worse, its copy constructor), but if std::move is > > eliminated, the return statement becomes effectively free, because it's > an > > elidable copy/move. In this case it's not at all clear to me how to fix > the > > problem; it seems like a fix would either require special-casing > std::move > > (which seems like a hack, and breaks the convention that the > core-language > > standard tries not to refer to the library standard), or require the > > implementation to 'see into' the implementation of a function called in > a > > return statement (which seems impractical). > > > > I think it's worth trying to fix these issues, despite the difficulties, > > because being able to provide reliable "rules of thumb" would > substantially > > improve the learning curve for std::move. Can anyone suggest how these > > issues could be fixed? > > Per the above, I think we can make "you can leave out std::move when > returning a named local variable" be the right advice, with the added > bonus of a compiler warning if you put in a pessimizing std::move. > -- --- 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 email 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-proposals/. ------=_Part_296_23524625.1372104509838 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable I remember sending you and Mike a reminder about that some months ago, and = was wondering, did you actually talk about this at the Bristol meeting?
=
On Monday, June 24, 2013 9:14:17 PM UTC+2, Richard Smith wrote:On Mon, Jun 24, 2013 at 11:42 AM, Geof= frey Romer <gro...@google.com> wrote:
> It's surprisingly hard to come up with a good, teachable set of ru= les for
> when to use std::move. For example, it would be nice to be able to= say "you
> can leave out std::move when returning a named local variable", bu= t this is
> not always the case. Consider:
>
> std::unique_ptr<const Foo> FooFactory() {
>   std::unique_ptr<Foo> result(new Foo);
>   // ...
>   return std::move(result);
> }
>
> The "std::move" is mandatory here, because an lvalue can only be i= mplicitly
> treated as an rvalue during copy operations, and in "return result= ;",
> 'result' is not copied, but used as the input for an implicit conv= ersion. It
> seems like this should be fixable, perhaps by expanding the implic= it-rvalue
> rules, but I'm not sure what all the consequences would be (at a m= inimum,
> this could change the behavior of existing code that has separate = rvalue and
> lvalue overloads for an implicit conversion operation).

Yes, for NRVO, I think the "same cv-unqualified type as the function
return type" restriction only needs to apply to copy-elision and not
to implicit move. That is, I think we should treat 'result' as an
xvalue here:

T f() {
  U result;
  return result;
}

... independent of the types of T and U, but we should only be
permitted to perform copy-elision of they're the same cv-unqualified
type. Most generalizations of the implicit-rvalue rules are fraught
with peril, but this one seems relatively safe.

> Conversely, it would be nice to be able to say "When in doubt, jus= t use
> std::move if you want move semantics, because it never hurts", but= std::move
> can in fact be a pessimization:
>
> std::array<Foo, 10000> MakeHugeArray() {
>   std::array<Foo, 10000> result;
>   // ...
>   return std::move(result);
> }

I suggest you talk to your compiler vendor (Hi!) and ask for a warning
for this case.

> As written, the return statement incurs 10,000 invocations of Foo'= s move
> constructor (or, worse, its copy constructor), but if std::move is
> eliminated, the return statement becomes effectively free, because= it's an
> elidable copy/move. In this case it's not at all clear to me how t= o fix the
> problem; it seems like a fix would either require special-casing s= td::move
> (which seems like a hack, and breaks the convention that the core-= language
> standard tries not to refer to the library standard), or require t= he
> implementation to 'see into' the implementation of a function call= ed in a
> return statement (which seems impractical).
>
> I think it's worth trying to fix these issues, despite the difficu= lties,
> because being able to provide reliable "rules of thumb" would subs= tantially
> improve the learning curve for std::move. Can anyone suggest how t= hese
> issues could be fixed?

Per the above, I think we can make "you can leave out std::move when
returning a named local variable" be the right advice, with the added
bonus of a compiler warning if you put in a pessimizing std::move.

--
 
---
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-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-proposals/<= /a>.
 
 
------=_Part_296_23524625.1372104509838-- .