www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Encapsulate return value in scoped

reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
Is there a way to encapsulate return value into scoped?

Say I have a function that returns a new object:

X new_x(T t...) {
     //Super complex input processing
     return new X(something);
}

And I want to encapsulate the result using scoped, is that 
possible? Can I just do:
     return scoped!X(something).
?

If I understand correctly, the data will be blitted, but the 
destructor will be called, so the returned object will be in 
invalid state.
Jun 11 2015
parent reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
On Thursday, 11 June 2015 at 08:48:22 UTC, Yuxuan Shui wrote:

 Is there a way to encapsulate return value into scoped?

 Say I have a function that returns a new object:

 X new_x(T t...) {
     //Super complex input processing
     return new X(something);
 }

 And I want to encapsulate the result using scoped, is that 
 possible? Can I just do:
     return scoped!X(something).
 ?

 If I understand correctly, the data will be blitted, but the 
 destructor will be called, so the returned object will be in 
 invalid state.


It's even weirder than I thought, this:

import std.stdio,
        std.typecons;
class A {
	int b;
	~this() {
		writeln("Des");
	}
	this(int x) {
		b=x;
		writeln("Cons");
	}
}
auto x() {
	A x = scoped!A(10);
	writeln(x.b);
	writeln("Return x");
	return x;
}
void main() {
	auto tx = x();
	writeln("Return main");
}

Produce output:
Cons
Des
0
Return x
Return main

Which I totally don't understand.
Jun 11 2015
parent reply Daniel =?UTF-8?B?S296w6Fr?= via Digitalmars-d-learn writes:
On Thu, 11 Jun 2015 09:01:04 +0000
Yuxuan Shui via Digitalmars-d-learn <digitalmars-d-learn puremagic.com>
wrote:

 	A x = scoped!A(10);


use auto x = scoped!A(10);
Jun 11 2015
parent reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
On Thursday, 11 June 2015 at 09:11:47 UTC, Daniel Kozák wrote:

 On Thu, 11 Jun 2015 09:01:04 +0000
 Yuxuan Shui via Digitalmars-d-learn 
 <digitalmars-d-learn puremagic.com>
 wrote:

 	A x = scoped!A(10);


 use auto x = scoped!A(10);


Thanks!

Curious question, why doesn't compiler reject this code?
Jun 11 2015
parent reply Steven Schveighoffer <schveiguy yahoo.com> writes:
On 6/11/15 1:28 PM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 09:11:47 UTC, Daniel Kozák wrote:

 On Thu, 11 Jun 2015 09:01:04 +0000
 Yuxuan Shui via Digitalmars-d-learn <digitalmars-d-learn puremagic.com>
 wrote:

     A x = scoped!A(10);


 use auto x = scoped!A(10);


 Thanks!

 Curious question, why doesn't compiler reject this code?


Because scoped!A implicitly casts to A.

-Steve
Jun 11 2015
parent reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
On Thursday, 11 June 2015 at 17:34:56 UTC, Steven Schveighoffer 
wrote:

 On 6/11/15 1:28 PM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 09:11:47 UTC, Daniel Kozák wrote:

 On Thu, 11 Jun 2015 09:01:04 +0000
 Yuxuan Shui via Digitalmars-d-learn 
 <digitalmars-d-learn puremagic.com>
 wrote:

    A x = scoped!A(10);


 use auto x = scoped!A(10);


 Thanks!

 Curious question, why doesn't compiler reject this code?


 Because scoped!A implicitly casts to A.

 -Steve


Thanks!

I just found that out myself. Learned 'alias this' in the process.
Jun 11 2015
parent reply =?UTF-8?B?QWxpIMOHZWhyZWxp?= <acehreli yahoo.com> writes:
On 06/11/2015 11:43 AM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 17:34:56 UTC, Steven Schveighoffer wrote:

 On 6/11/15 1:28 PM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 09:11:47 UTC, Daniel Kozák wrote:

 On Thu, 11 Jun 2015 09:01:04 +0000
 Yuxuan Shui via Digitalmars-d-learn <digitalmars-d-learn puremagic.com>
 wrote:

    A x = scoped!A(10);


 use auto x = scoped!A(10);


 Thanks!

 Curious question, why doesn't compiler reject this code?


 Because scoped!A implicitly casts to A.

 -Steve


 Thanks!

 I just found that out myself. Learned 'alias this' in the process.


Shameless plug: :)

   http://ddili.org/ders/d.en/destroy.html#ix_destroy.scoped

This issue is explained at the end of that section.

Ali
Jun 11 2015
parent reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
On Thursday, 11 June 2015 at 19:23:49 UTC, Ali Çehreli wrote:

 On 06/11/2015 11:43 AM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 17:34:56 UTC, Steven 
 Schveighoffer wrote:

 On 6/11/15 1:28 PM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 09:11:47 UTC, Daniel Kozák 
 wrote:

 On Thu, 11 Jun 2015 09:01:04 +0000
 Yuxuan Shui via Digitalmars-d-learn 
 <digitalmars-d-learn puremagic.com>
 wrote:

   A x = scoped!A(10);


 use auto x = scoped!A(10);


 Thanks!

 Curious question, why doesn't compiler reject this code?


 Because scoped!A implicitly casts to A.

 -Steve


 Thanks!

 I just found that out myself. Learned 'alias this' in the 
 process.


 Shameless plug: :)

   http://ddili.org/ders/d.en/destroy.html#ix_destroy.scoped

 This issue is explained at the end of that section.

 Ali


Can you explain more about why the destructor is not called when 
returning a struct?

Can't seem to find it in the document.
Jun 11 2015
parent reply =?UTF-8?B?QWxpIMOHZWhyZWxp?= <acehreli yahoo.com> writes:
On 06/11/2015 12:51 PM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 19:23:49 UTC, Ali Çehreli wrote:



   http://ddili.org/ders/d.en/destroy.html#ix_destroy.scoped

 This issue is explained at the end of that section.

 Ali


 Can you explain more about why the destructor is not called when
 returning a struct?


Are you asking in general or specific to scoped!C?

In general, D has move semantics built into the language. It depends on 
whether the returned expression is an rvalue or an lvalue: rvalues are 
moved, lvalues are copied. And the destructor will not be called for a 
moved object.

About returning scoped!C, I think it works:

import std.stdio;
import std.typecons;

class C
{
     ~this()
     {
         writeln("dtor");
     }
}

auto foo()
{
     auto c = scoped!C();
     return c;
}

void main()
{
     writeln("entering scope");
     {
         writeln("calling");
         auto s = foo();
         writeln("returned");
     }
     writeln("leaving scope");
}

"dtor" is printed upon leaving the scope:

entering scope
calling
returned
dtor
leaving scope

Ali
Jun 11 2015
parent reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
On Thursday, 11 June 2015 at 21:38:59 UTC, Ali Çehreli wrote:

 On 06/11/2015 12:51 PM, Yuxuan Shui wrote:

 On Thursday, 11 June 2015 at 19:23:49 UTC, Ali Çehreli wrote:



 [...]


 Can you explain more about why the destructor is not called


 when

 returning a struct?


 Are you asking in general or specific to scoped!C?

 In general, D has move semantics built into the language. It 
 depends on whether the returned expression is an rvalue or an 
 lvalue: rvalues are moved, lvalues are copied. And the 
 destructor will not be called for a moved object.

 About returning scoped!C, I think it works:

 import std.stdio;
 import std.typecons;

 class C
 {
     ~this()
     {
         writeln("dtor");
     }
 }

 auto foo()
 {
     auto c = scoped!C();
     return c;
 }

 void main()
 {
     writeln("entering scope");
     {
         writeln("calling");
         auto s = foo();
         writeln("returned");
     }
     writeln("leaving scope");
 }

 "dtor" is printed upon leaving the scope:

 entering scope
 calling
 returned
 dtor
 leaving scope

 Ali


I just find out that the document of scoped says that "It's 
illegal to move a class instance even if you are sure there are 
no pointers to it. As such, it is illegal to move a scoped 
object."

So this is not a solution?
Jun 18 2015
parent reply =?UTF-8?B?QWxpIMOHZWhyZWxp?= <acehreli yahoo.com> writes:
On 06/18/2015 04:53 PM, Yuxuan Shui wrote:


 On Thursday, 11 June 2015 at 21:38:59 UTC, Ali Çehreli wrote:



 About returning scoped!C, I think it works:





 I just find out that the document of scoped says that "It's


Thanks for fixing the typo there. (The documentations has lower case: 
"it's". ;) )


 illegal to
 move a class instance


Wait! You fixed that as well? :) The doc that I am looking at says 
"class reference":




 even if you are sure there are no pointers to it.
 As such, it is illegal to move a scoped object."

 So this is not a solution?


I guess not. :-/

Ali
Jun 18 2015
parent reply "Yuxuan Shui" <yshuiv7 gmail.com> writes:
On Friday, 19 June 2015 at 00:00:50 UTC, Ali Çehreli wrote:

 On 06/18/2015 04:53 PM, Yuxuan Shui wrote:


 On Thursday, 11 June 2015 at 21:38:59 UTC, Ali Çehreli wrote:



 About returning scoped!C, I think it works:





 I just find out that the document of scoped says that "It's


 Thanks for fixing the typo there. (The documentations has lower 
 case: "it's". ;) )


 illegal to
 move a class instance


 Wait! You fixed that as well? :) The doc that I am looking at 
 says "class reference":




Those are fixed by: 
https://github.com/D-Programming-Language/phobos/pull/3016


 even if you are sure there are no pointers to it.
 As such, it is illegal to move a scoped object."

 So this is not a solution?


 I guess not. :-/

 Ali


I don't understand. What problem can moving a scoped cause when 
there's no reference to its inner class?
Jun 18 2015
parent =?UTF-8?B?QWxpIMOHZWhyZWxp?= <acehreli yahoo.com> writes:
On 06/18/2015 05:25 PM, Yuxuan Shui wrote:


 What problem can moving a scoped cause when there's
 no reference to its inner class?


I am curious as well. Could it be related to the object's 'monitor'? 
Perhaps some other code may have a reference to monitor? And that 
reference does not constitute as a reference to the object itself? I 
don't know.

Ali
Jun 18 2015