Thread overview
Casting between Structs
Feb 08
jxel
February 08
I reduced an issue in my project to this code snippet:

struct Struct1 { }
struct Struct2 { }

void func1(const Struct1);
void func2()
{
    Struct2 s;
    func1(cast(Struct1)s);
}


DMD compiles this just fine but according to the spec this should not compile. The spec (https://dlang.org/spec/expression.html#CastExpression section 8) says that "Casting a value v to a struct S, when value is not a struct of the same type, is equivalent to: S(v)", however, if I write Struct1(s) in this example, I get a compile error.

Is the spec wrong or is this a problem in the DMD compiler?

P.S. this snippet causes LDC to segfault (https://github.com/ldc-developers/ldc/issues/3314), but it's not clear whether it should be made to work like DMD, or if they should conform to the spec and make this a compile error
February 08
On Saturday, 8 February 2020 at 17:36:17 UTC, Jonathan Marler wrote:
> I reduced an issue in my project to this code snippet:
>
> struct Struct1 { }
> struct Struct2 { }
>
> void func1(const Struct1);
> void func2()
> {
>     Struct2 s;
>     func1(cast(Struct1)s);
> }
>
>
> DMD compiles this just fine but according to the spec this should not compile. The spec (https://dlang.org/spec/expression.html#CastExpression section 8) says that "Casting a value v to a struct S, when value is not a struct of the same type, is equivalent to: S(v)", however, if I write Struct1(s) in this example, I get a compile error.
>
> Is the spec wrong or is this a problem in the DMD compiler?
>
> P.S. this snippet causes LDC to segfault (https://github.com/ldc-developers/ldc/issues/3314), but it's not clear whether it should be made to work like DMD, or if they should conform to the spec and make this a compile error

The spec is wrong in many aspects. It was written then never followed. Not sure if it even gets updated.

This seems like it should be an error. I don't know why you would ever want to do a cast like that, especially with lvalues. And if you did want to there is an easy alternative *cast(S1*)&v.
February 09
On Saturday, 8 February 2020 at 19:28:26 UTC, jxel wrote:
> On Saturday, 8 February 2020 at 17:36:17 UTC, Jonathan Marler wrote:
>> [...]
>
> The spec is wrong in many aspects. It was written then never followed. Not sure if it even gets updated.
>

I hope you're wrong about this.  Can someone in D leadership please prove this statement wrong by clarifying whether this is a bug in DMD or a missing feature in the spec?


February 09
On Sunday, 9 February 2020 at 01:54:36 UTC, Jonathan Marler wrote:
> On Saturday, 8 February 2020 at 19:28:26 UTC, jxel wrote:
>> On Saturday, 8 February 2020 at 17:36:17 UTC, Jonathan Marler wrote:
>>> [...]
>>
>> The spec is wrong in many aspects. It was written then never followed. Not sure if it even gets updated.
>>
>
> I hope you're wrong about this.  Can someone in D leadership please prove this statement wrong by clarifying whether this is a bug in DMD or a missing feature in the spec?

I've created a bug here: "Struct Cast Not Documented or Incorrectly Accepted" https://issues.dlang.org/show_bug.cgi?id=20570
February 13
On 2/8/2020 6:05 PM, Jonathan Marler wrote:
> I've created a bug here: "Struct Cast Not Documented or Incorrectly Accepted" https://issues.dlang.org/show_bug.cgi?id=20570

It shouldn't compile. I'd have to investigate with the implementation.