https://d.puremagic.com/issues/show_bug.cgi?id=12024



--- Comment #11 from Vladimir Panteleev <thecybershadow@gmail.com> 2014-01-29 16:07:02 EET ---
Derp, unit test line should be:

    struct S3{union{int m;const int c;}}

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024



--- Comment #12 from monarchdodra@gmail.com 2014-01-29 06:26:10 PST ---
(In reply to comment #8)
> It should be possibly by comparing .offsetof with other members.

Smart. Unfortunatly, for a "recursive" implementation, it is a bit difficult to exploit: More often than not, you want to know before hand that you are about to process an aggregate in an union.

(In reply to comment #9)
> If a non-mutable field has one or more overlapped union _mutable_ fields, the whole struct is treated as modifiable.
> 
> import std.traits : Unqual;
> struct Rebindable(T)
> {
>     union
>     {
>         T origin;
>         Unqual!T stripped;  // overlapped union mutable field of 'origin'
>     }
> }

Yes, absolutely. That's what I had in mind. But as I said, even detecting that you are in an (anonymous) union is a bit difficult. Do-able with your suggestion, just... difficult.

(In reply to comment #10)
> Interesting. So this should "just work":
> 
> diff --git a/std/algorithm.d b/std/algorithm.d
> index 036b918..0ed5606 100644
> --- a/std/algorithm.d
> +++ b/std/algorithm.d
> @@ -2054,6 +2054,9 @@ void swap(T)(ref T lhs, ref T rhs) if
> (is(typeof(lhs.proxySwap(rhs))))
>  private template allMutableFields(T)
>  {
>      alias OT = OriginalType!T;
> +    static if (is(typeof({ T t = void; t = t; })))
> +        enum allMutableFields = true;
> +    else
>      static if (is(OT == struct) || is(OT == union))
>          enum allMutableFields = isMutable!OT && allSatisfy!(.allMutableFields,
> FieldTypeTuple!OT);
>      else
> @@ -2072,6 +2075,9 @@ unittest
>      struct S2{const int i;}
>      static assert( allMutableFields!S1);
>      static assert(!allMutableFields!S2);
> +
> +    struct S3{union X{int m;const int c;}X x;}
> +    static assert( allMutableFields!S3);
>  }

I think that's wrong, because "static if (is(typeof({ T t = void; t = t; })))" can work in the pressence of an opAssign: Complex structs with immutable members but with a valid opAssign can't be swapped. The only case where it would work is if T didn't have an opAssign to begin with. However, that test would be mostly useless, since we'd have to deploy code for structs that do have opAssign regardless.

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024



--- Comment #13 from Kenji Hara <k.hara.pg@gmail.com> 2014-01-29 06:36:51 PST ---
(In reply to comment #12)
> (In reply to comment #8)
> > It should be possibly by comparing .offsetof with other members.
> 
> Smart. Unfortunatly, for a "recursive" implementation, it is a bit difficult to exploit: More often than not, you want to know before hand that you are about to process an aggregate in an union.
> 
> (In reply to comment #9)
> > If a non-mutable field has one or more overlapped union _mutable_ fields, the whole struct is treated as modifiable.
> > 
> > import std.traits : Unqual;
> > struct Rebindable(T)
> > {
> >     union
> >     {
> >         T origin;
> >         Unqual!T stripped;  // overlapped union mutable field of 'origin'
> >     }
> > }
> 
> Yes, absolutely. That's what I had in mind. But as I said, even detecting that you are in an (anonymous) union is a bit difficult. Do-able with your suggestion, just... difficult.

I think adding std.traits.isBlitAssignable(T) for the purpose would be good. And you can refer the code in the compiler.

https://github.com/D-Programming-Language/dmd/blob/master/src/mtype.c#L8588 int TypeStruct::isAssignable()

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024


Kenji Hara <k.hara.pg@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |pull, rejects-valid


--- Comment #14 from Kenji Hara <k.hara.pg@gmail.com> 2014-01-30 01:55:30 PST ---
https://github.com/D-Programming-Language/phobos/pull/1891

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024



--- Comment #15 from github-bugzilla@puremagic.com 2014-01-31 03:28:55 PST ---
Commits pushed to master at https://github.com/D-Programming-Language/phobos

https://github.com/D-Programming-Language/phobos/commit/2d32c78af43a3d0a1b9015a8c971cefed9a1035f fix Issue 12024 - template instantiation for swap(SysTime, SysTime) fails

https://github.com/D-Programming-Language/phobos/commit/b8f242e78f587f41aa344173b8961bf613e20c0d Merge pull request #1891 from 9rnsr/fix12024

[REG2.065a] Issue 12024 - template instantiation for swap(SysTime, SysTime)
fails

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024


Kenji Hara <k.hara.pg@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024



--- Comment #16 from github-bugzilla@puremagic.com 2014-02-01 04:51:01 PST ---
Commit pushed to 2.065 at https://github.com/D-Programming-Language/phobos

https://github.com/D-Programming-Language/phobos/commit/84422029eb26ccea2cd97422c2e731792ae1e114 Merge pull request #1891 from 9rnsr/fix12024

[REG2.065a] Issue 12024 - template instantiation for swap(SysTime, SysTime)
fails
Conflicts:
    std/algorithm.d

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

https://d.puremagic.com/issues/show_bug.cgi?id=12024



--- Comment #17 from github-bugzilla@puremagic.com 2014-02-27 20:13:09 PST ---
Commit pushed to master at https://github.com/D-Programming-Language/phobos

https://github.com/D-Programming-Language/phobos/commit/84422029eb26ccea2cd97422c2e731792ae1e114 Merge pull request #1891 from 9rnsr/fix12024

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------