220 6938 article Path: news.gmane.org!not-for-mail From: Andrew Tomazos Newsgroups: gmane.comp.lang.c++.isocpp.proposals Subject: Re: constexpr-specified loop variable Date: Thu, 3 Oct 2013 15:56:23 -0700 (PDT) Lines: 150 Approved: news@gmane.org Message-ID: References: <8527db17-8c8a-43c5-b159-4e14c879f774@isocpp.org> Reply-To: std-proposals@isocpp.org NNTP-Posting-Host: plane.gmane.org Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_48_5406402.1380840983781" X-Trace: ger.gmane.org 1380840982 20092 80.91.229.3 (3 Oct 2013 22:56:22 GMT) X-Complaints-To: usenet@ger.gmane.org NNTP-Posting-Date: Thu, 3 Oct 2013 22:56:22 +0000 (UTC) To: std-proposals@isocpp.org Original-X-From: std-proposals+bncBD5KHQXXWYPRBGHMW6JAKGQEBDVQ3HI@isocpp.org Fri Oct 04 00:56:27 2013 Return-path: Envelope-to: gclcip-std-proposals@m.gmane.org Original-Received: from mail-yh0-f70.google.com ([209.85.213.70]) by plane.gmane.org with esmtp (Exim 4.69) (envelope-from ) id 1VRron-0007HC-Nw for gclcip-std-proposals@m.gmane.org; Fri, 04 Oct 2013 00:56:26 +0200 Original-Received: by mail-yh0-f70.google.com with SMTP id i72sf6811289yha.5 for ; Thu, 03 Oct 2013 15:56:24 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=date:from:to:message-id:in-reply-to:references:subject:mime-version :x-original-sender:reply-to:precedence:mailing-list:list-id :list-post:list-help:list-archive:list-subscribe:list-unsubscribe :content-type; bh=BdShtsZnduUQ8L5t+PASXOhPSOoy4IoyMvt/BkUk0vY=; b=UWB5sh57xHTIA7d0E5yHKJKK7397iikXn4e3mMYCvDYRkLV604PphoVu8lmmq5wBgV YoUgI48RVhiJq3S2Y0KVEFSFtsHGBVXxv6d7SGn1EzWV5PNUFNQKpEVwZD96RQBfBfAx RvEru3rP7f5oF0ZVzyK+6X4FLpyMSh+bgeJpzSiuHsKVk6VAq4fbu8J2WnNQDHr3Qj0t C4p7XoeubRnl5mDpwOaVbfs1RW9CEiJVtXa5oVDrFd4/AGauDLD5x7cTKMP7xvmnpihM l/UttLvqsnOIWnp+Bzu2F9FeGHl14bPhSraA3znnSP+Nm7w5oYI0rjm17sAJhLkZK2ys kSBA== X-Received: by 10.236.13.36 with SMTP id a24mr9389812yha.50.1380840984836; Thu, 03 Oct 2013 15:56:24 -0700 (PDT) X-BeenThere: std-proposals@isocpp.org Original-Received: by 10.49.0.12 with SMTP id 12ls1141066qea.25.gmail; Thu, 03 Oct 2013 15:56:24 -0700 (PDT) X-Received: by 10.49.133.225 with SMTP id pf1mr454172qeb.6.1380840984255; Thu, 03 Oct 2013 15:56:24 -0700 (PDT) In-Reply-To: X-Original-Sender: andrewtomazos@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:6938 Archived-At: ------=_Part_48_5406402.1380840983781 Content-Type: text/plain; charset=ISO-8859-1 On Thursday, October 3, 2013 10:34:12 PM UTC+2, Richard Smith wrote: > > It's not true that we can't use 'constexpr' to specify a loop variable. > Yes, as per my followup to Daniel I meant that it isn't effectively possible because of how range-based for is currently defined. The semantics are as if the loop is unrolled in the obvious way. >> > > If you want to force loop unrolling, it would seem much more natural to > use a pack expansion. > The purpose isn't to force unrolling, the term "unrolled" is purely for specification purposes. An implementation is free to unroll a normal loop, and is free to rollup an "unrolled" loop. It's purely the logical behaviour of the program that is specified. Using a construct like 'for' would violate peoples' intuitions. 'for' does > not usually imply code duplication, > The behaviour of a program containing: for (T x : L) and the same one instead containing: for (constexpr T x : L) are identical if both well-formed. So it isn't clear how it is unintuitive. As stated the code duplication is only logical. a 'constexpr' variable declaration can usually have only a single value > In the following: template constexpr int g() { constexpr int y = x*2; return y + 5; } constexpr int z1 = g<3>(); constexpr int z2 = g<4>(); The `y` constexpr variable declaration has two different values, 6 and 8. Just like in the following: for (constexpr int x : {3,4}) { constexpr int y = x*2; /* ... */ } We can think of the inner statement of the unrolled for statement as if it were the body of a unnamed template function, and the loop variable as a non-type template parameter. The body is instantiated and called once for each element of the constexpr array initializer. -- --- 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_48_5406402.1380840983781 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
On Thursday, October 3, 2013 10:34:12 PM UTC+2, Richard Sm= ith wrote:
It's not true that we can't use 'constexpr' to specify a loop variable.

Yes, as per my follow= up to Daniel I meant that it isn't effectively possible because of how rang= e-based for is currently defined.

The= semantics are as if the loop is unrolled in the obvious way.

If you want to force loop unrolling,= it would seem much more natural to use a pack expansion.
 
The purpose isn't to force unrolling, the term "unrolled"= is purely for specification purposes.  An implementation is free to u= nroll a normal loop, and is free to rollup an "unrolled" loop.  It's p= urely the logical behaviour of the program that is specified.

Using a construct = like 'for' would violate peoples' intuitions.  'for'=20 does not usually imply code duplication,
 The behaviour of a program containing:

    &nbs= p; for (T x : L)

and the same one instead containing:

 &= nbsp;  for (constexpr T x : L)

are identical if both well-forme= d.  So it isn't clear how it is unintuitive.  As stated the code = duplication is only logical.

a 'constexpr' variable=20 declaration can usually have only a single value

In the following:

    template<int x>
   = ; constexpr int g()
    {
        constexpr int y = =3D x*2;

  &nbs= p;     return y + 5;
    }

    constexpr int z= 1 =3D g<3>();
    constexpr int z2 =3D g<4>()= ;

The `y` constexpr variable declaration has two different va= lues, 6 and 8.

Just like in the following:

    for (constexpr int x := {3,4})
    {
      &nbs= p; constexpr int y =3D x*2;

      &nbs= p; /* ... */
    }

We can think of the inne= r statement of the unrolled for statement as if it were the body of a unnam= ed template function, and the loop variable as a non-type template paramete= r.  The body is instantiated and called once for each element of the c= onstexpr array initializer.

--
 
---
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_48_5406402.1380840983781-- .