Bug 11270 – [REG 2.064] Initialization of struct in constructor

The following code used to work dmd 2.064 beta 1 and all versions before that. It just broke on the latest version of the 2.064 branch on git. repro.d(13): Error: cannot implicitly convert expression (f) of type float to netvar!float Although I'm not really sure if this is a bug or if the code is invalid and just used to work because of other bugs. Repro case: struct netvar(T) { T value; alias value this; void opAssign(T newVal){ value = newVal; } } void main(string[] args) { float f = 2.0f; netvar!float f2 = f; }

(In reply to comment #0) > The following code used to work dmd 2.064 beta 1 and all versions before that. > It just broke on the latest version of the 2.064 branch on git. > > repro.d(13): Error: cannot implicitly convert expression (f) of type float to > netvar!float With 2.057-2.063, same error occurs for the code. > Although I'm not really sure if this is a bug or if the code is invalid and > just used to work because of other bugs. As far as I know, the repro case is invalid. On the variable initialization `netvar!float f2 = f;` opAssign never invoked.

Sorry, I reduced the repro case to much. Here is the actual repro case I had first: struct netvar(T) { T value; alias value this; void opAssign(T newVal){ value = newVal; } } class Foo { netvar!float f; this(float f) { this.f = f; } } void main(string[] args) { new Foo(2.0f); }

(In reply to comment #2) > Sorry, I reduced the repro case to much. Here is the actual repro case I had > first: > > struct netvar(T) { > T value; > alias value this; > > void opAssign(T newVal){ > value = newVal; > } > } > > class Foo > { > netvar!float f; > > this(float f) > { > this.f = f; > } > } > > void main(string[] args) > { > new Foo(2.0f); > } Ah, this is an expected behavior that has been introduced by fixing bug 9665. With git-head, compiler tries to rewrite `this.f = f;` to the field initialization. Possible modifications: 1. Rewrite this.f = f; to this.f = netvar!float(f); It will initialize the Foo.f by the struct literal. 2. Add a constructor this(T); in netbar struct to accept a T value as a valid initializer.

Thanks for the suggestions. I knew how to fix this, but I just wanted to confirm that this is indeed intended behavior and not a regression, because it used to work in all earlier compiler versions.

(In reply to comment #4) > Thanks for the suggestions. I knew how to fix this, but I just wanted to > confirm that this is indeed intended behavior and not a regression, because it > used to work in all earlier compiler versions. It was the long standing bug. I'm glad to be able to fix it finally. Mark this issue as resolved invalid.

*** Issue 11458 has been marked as a duplicate of this issue. ***