Bug 8950 – postblit not called on const static array initialization

--- struct S { static int postblits = 0; this(this) const { ++postblits; } } void main() { auto s = S(); S[2] arr = s; assert(S.postblits == 2); // ok S.postblits = 0; const S[2] constArr = s; assert(S.postblits == 2); // failed: S.postblits == 0 S.postblits = 0; const S[2] constArr2 = arr; assert(S.postblits == 2); // failed: S.postblits == 0 S.postblits = 0; } ---

The behavior of this has changed a bit. First, qualified postblits are deprecated. Taking away the qualified, the second test passes, but not the third one. However postblit is going to be deprecated in the future so I doubt this'll get fixed. Leaving open until the postblit deprecation though...

Yes, this will get a WONTFIX.

Copy ctors are currently (wrongly) ignored when copying static arrays: https://issues.dlang.org/show_bug.cgi?id=20365 This issue here is pretty bad and inconsistent - the postblit is called correctly for mutable copies and scalar non-mutable copies, only non-mutable static array copies ignore the postblit: void main() { static struct S { int x = 42; this(this) { x += 10; } } { S source; S mutableCopy = source; assert(mutableCopy.x == 52); const S constCopy = source; assert(constCopy.x == 52); } { S[1] source; auto mutableCopy = source; assert(mutableCopy[0].x == 52); const constCopy = source; assert(constCopy[0].x == 52); // fails immutable immutableCopy = source; assert(immutableCopy[0].x == 52); // fails } }

Issue 20365 has been fixed and kinke's code and the original provided code do not assert anymore. Closing as WORKSFORME.