220 6988 article Path: news.gmane.org!not-for-mail From: Richard Smith Newsgroups: gmane.comp.lang.c++.isocpp.proposals Subject: Re: Unnamed LValues Date: Fri, 4 Oct 2013 18:33:43 -0700 Lines: 213 Approved: news@gmane.org Message-ID: References: Reply-To: std-proposals@isocpp.org NNTP-Posting-Host: plane.gmane.org Mime-Version: 1.0 Content-Type: multipart/alternative; boundary=047d7bf0f67ad78a0804e7f468d5 X-Trace: ger.gmane.org 1380936824 31254 80.91.229.3 (5 Oct 2013 01:33:44 GMT) X-Complaints-To: usenet@ger.gmane.org NNTP-Posting-Date: Sat, 5 Oct 2013 01:33:44 +0000 (UTC) To: std-proposals@isocpp.org Original-X-From: std-proposals+bncBDVNBJG4YAIBB6GYXWJAKGQEOCGPGAI@isocpp.org Sat Oct 05 03:33:48 2013 Return-path: Envelope-to: gclcip-std-proposals@m.gmane.org Original-Received: from mail-qa0-f72.google.com ([209.85.216.72]) by plane.gmane.org with esmtp (Exim 4.69) (envelope-from ) id 1VSGkb-0002BD-EW for gclcip-std-proposals@m.gmane.org; Sat, 05 Oct 2013 03:33:45 +0200 Original-Received: by mail-qa0-f72.google.com with SMTP id j7sf4098293qaq.11 for ; Fri, 04 Oct 2013 18:33:44 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from: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=zirIVQ8i8mnEDMthO8sdcjhuNvwwVtX816LXjPRgf1Q=; b=WBGJ4wYi1UH0Hrq1Ao8cB38LuGwsun5EGo7rTWC1DQvJVR1FhLJZiQwTJZgXU4kznr GF+xnJKfNZqT4G/FutQXdsqcX2goO3zRdyR8QY/F41iIbNB4IGgjz8gGYROpdBKfAnwO L8gmzSXE3wfo4kui8XYrSCpyQgJ/8WqU0ySg+K74feMG6ezt8zx1G4rdo89FS6XNEa9b 1Wpz6qzRrq2bLiOvVcA8R0+BlhsYKw6ykbRULhFyZKsYFtFdT6IOfr5AsmV7EETqhJCo VsUFUz3AxmX4kB/lfVuCUkVgsslFC8xePH62aAu3fmEhccV4+ttk9NS6vdoMSR25w/30 dm+w== X-Received: by 10.236.56.70 with SMTP id l46mr17510319yhc.2.1380936824599; Fri, 04 Oct 2013 18:33:44 -0700 (PDT) X-BeenThere: std-proposals@isocpp.org Original-Received: by 10.49.104.84 with SMTP id gc20ls1565478qeb.96.gmail; Fri, 04 Oct 2013 18:33:44 -0700 (PDT) X-Received: by 10.220.145.132 with SMTP id d4mr14759221vcv.9.1380936823962; Fri, 04 Oct 2013 18:33:43 -0700 (PDT) Original-Received: from mail-vb0-x22c.google.com (mail-vb0-x22c.google.com [2607:f8b0:400c:c02::22c]) by mx.google.com with ESMTPS id nr7si4191602vec.27.1969.12.31.16.00.00 (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Fri, 04 Oct 2013 18:33:43 -0700 (PDT) Received-SPF: pass (google.com: domain of metafoo@gmail.com designates 2607:f8b0:400c:c02::22c as permitted sender) client-ip=2607:f8b0:400c:c02::22c; Original-Received: by mail-vb0-f44.google.com with SMTP id e13so2608533vbg.17 for ; Fri, 04 Oct 2013 18:33:43 -0700 (PDT) X-Received: by 10.220.181.136 with SMTP id by8mr15047029vcb.11.1380936823833; Fri, 04 Oct 2013 18:33:43 -0700 (PDT) Original-Sender: metafoo@gmail.com Original-Received: by 10.220.244.137 with HTTP; Fri, 4 Oct 2013 18:33:43 -0700 (PDT) In-Reply-To: X-Original-Sender: richard@metafoo.co.uk X-Original-Authentication-Results: mx.google.com; spf=pass (google.com: domain of metafoo@gmail.com designates 2607:f8b0:400c:c02::22c as permitted sender) smtp.mail=metafoo@gmail.com; dkim=pass header.i=@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:6988 Archived-At: --047d7bf0f67ad78a0804e7f468d5 Content-Type: text/plain; charset=ISO-8859-1 On Fri, Oct 4, 2013 at 6:01 PM, Brent Friedman wrote: > When using RAII, the need to give the variable an explicit name is > cumbersome and often not useful. I'll use the example of freezing and > thawing in WxWidgets, but the problem is applicable to most any RAII > scenario. > > class WxScopeFreeze; //Freeze() in constructor, Thaw() in destructor > > WxScopeFreeze Freeze(this); > > Later on, I want to freeze an additional widget. I should probably update > and maintain these variable names now. > > WxScopeFreeze FreezeThis(this); > WxScopeFreeze FreezeOther(Other); > > Unfortunately, omitting the name is no good here. Without a name, the > object lifetime ends at the end of the full expression and we don't get our > scope freezing behavior. > > WxScopeFreeze(this); //doesn't live for the full scope > > If your object is generated as the result of a function call (esp. for > template argument deduction) then the syntax grows yet still. > > auto&& FreezeThis = ScopeFreeze(this); > auto&& FreezeOther = ScopeFreeze(Other); > > > Let's consider ways to get around this issue. The obvious one would be a > macro. > #define SCOPE_FREEZE(expr) WxScopeFreeze Freeze##__LINE__##(expr); > > This hides the noise to some extent; you no longer have to think about > what to name this RAII object and there won't be any name clashes as long > as you don't declare multiples on the same line of code. The __COUNTER__ > extension supported by some compilers would give you some alternative. > __COUNTER__ would be quite unsafe to use in the name of a variable which > you declare in a header though. > > There are more downsides to the macro approach. > * It requires the existence of the macro for every RAII type helper you > want. > * For our safer __LINE__ solution, names can still clash if they were > defined in a separate file (eg, header). These may not generate compilation > errors, but could generate spurious static analysis warnings and confusion > when debugging. > * In a debugger window, you'll see variable names that you never created > and that may be difficult to refer back to. > * Its a macro > > Instead of macro invention, how about adding some syntax for unnamed > variables? > > You could repurpose auto when used as the name of a variable. > WxScopeFreeze auto(this); > WxScopeFreeze auto(Other); > auto auto{WxScopeFreeze(this)}; > > In that last example we use type deduction on an unnamed variable. It > would be nice to standardize that idea a bit more, but I'm not sure what > the syntax should be. > > > We could consider some other keyword to identify unnamed variables. > Unfortunately, the obvious syntax of providing no name is taken as I > mentioned. > > = is almost appealing, but wouldn't work well with lambdas. Ideally we can > bind logic to the lifetime of a lambda with unnamed variable syntax. > WxScopeFreeze =(this); > WxScopeFreeze ={Other}; > [ ==WxScopeFreeze(this) ](){}; //weird? > The important part in such a construct is the point at which the RAII object is destroyed. Perhaps the syntax should make that more explicit. How about: WxScopeFreeze(this) { // ... } // dtor invoked here. .... or ... /*template std::lock_guard locked(Mutex &);*/ locked(my_mutex) { } You can get that effect currently with: WxScopeFreeze(this), [&]{ // ... }(); .... although this use of the comma operator is unfamiliar to most, and this doesn't support control flow leaving the block. -- --- 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/. --047d7bf0f67ad78a0804e7f468d5 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
On Fri, Oct 4, 2013 at 6:01 PM, Brent Friedman <four= thgeek@gmail.com> wrote:
When using RAII, the need t= o give the variable an explicit name is cumbersome and often not useful. I&= #39;ll use the example of freezing and thawing in WxWidgets, but the proble= m is applicable to most any RAII scenario.

class WxScopeFreeze; //Freeze() in constructor, Thaw() in de= structor=A0

WxScopeFreeze Freeze(this);
=
Later on, I want to freeze an additional widget. I should pr= obably update and maintain these variable names now.

WxScopeFreeze FreezeThis(this);
WxScopeFreeze= FreezeOther(Other);

Unfortunately, omitting the n= ame is no good here. Without a name, the object lifetime ends at the end of= the full expression and we don't get our scope freezing behavior.

WxScopeFreeze(this); //doesn't live for the full sc= ope

If your object is generated as the result of a= function call (esp. for template argument deduction) then the syntax grows= yet still.

auto&& FreezeThis =3D ScopeFreeze(this);
<= div>auto&& FreezeOther =3D ScopeFreeze(Other);

=

Let's consider ways to get around this issue. The o= bvious one would be a macro.
#define SCOPE_FREEZE(expr) WxScopeFreeze Freeze##__LINE__##(expr);

This hides the noise to some extent; you no longer ha= ve to think about what to name this RAII object and there won't be any = name clashes as long as you don't declare multiples on the same line of= code. The __COUNTER__ extension supported by some compilers would give you= some alternative. __COUNTER__ would be quite unsafe to use in the name of = a variable which you declare in a header though.

There are more downsides to the macro approach.=A0
* It requires the existence of the macro for every RAII type hel= per you want.
* For our safer __LINE__ solution, names can still = clash if they were defined in a separate file (eg, header). These may not g= enerate compilation errors, but could generate spurious static analysis war= nings and confusion when debugging.
* In a debugger window, you'll see variable names that you n= ever created and that may be difficult to refer back to.
* Its a = macro

Instead of macro invention, how about adding= some syntax for unnamed variables?

You could repurpose auto when used as the name of a var= iable.
WxScopeFreeze auto(this);
WxScopeFreeze auto(Oth= er);
auto auto{WxScopeFreeze(this)};

In that last example we use type deduction on an unnamed variable. It would= be nice to standardize that idea a bit more, but I'm not sure what the= syntax should be.


We could conside= r some other keyword to identify unnamed variables. Unfortunately, the obvi= ous syntax of providing no name is taken as I mentioned.

=3D is almost appealing, but wouldn't work well wit= h lambdas. Ideally we can bind logic to the lifetime of a lambda with unnam= ed variable syntax.
WxScopeFreeze =3D(this);
WxScopeFre= eze =3D{Other};
[ =3D=3DWxScopeFreeze(this) ](){}; //weird?

The important part in such a construct is the point at wh= ich the RAII object is destroyed. Perhaps the syntax should make that more = explicit. How about:

=A0 WxScopeFreeze(this) {
=A0 =A0 // ...
=A0 } // dtor invoked here.

... or ...
=

=A0 /*template<typename Mutex> std::lock_guard<= ;Mutex> locked(Mutex &);*/

=A0 locked(my_mutex) {
=A0 }

You can get that effect currently with:

= =A0 WxScopeFreeze(this), [&]{
=A0 =A0 // ...
=A0 }(= );

... although this use of the comma operator is unfamiliar to= most, and this doesn't support control flow leaving the block.

--
 
---
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>.
--047d7bf0f67ad78a0804e7f468d5-- .