Bug 12339 – "alias this" not considered for qualified conversion

I'd expect an implicit "const S" => "S" conversion to work if an "alias this" call allows it. //------------ //alias T = int; //OK alias T = S; //FAIL static struct S { int* p; T asMutable() const; alias asMutable this; } void main() { immutable S s; T ss = s; } //------------ I see this as a needless limitation: The fact that the types match should not prevent the alias this call from happening.

I think it does not work because struct contains pointer. If you comment out it, it will work. Note that dispite you alias to S type, S itself contains indirections. //alias T = int; //OK //alias T = S; //FAIL alias T = SS; // OK static struct S { int* p; T asMutable() const { return T(); } alias asMutable this; } struct SS { } void main() { immutable S s; T ss = s; }

(In reply to comment #1) > I think it does not work because struct contains pointer. If you comment out > it, it will work. > > Note that dispite you alias to S type, S itself contains indirections. I know, that was the point. I *wanted* a type were "const(S)" can't be implicitly cast to "S" The "issue" is that "const(S)" has an alias this to a function that produces an "S", meaning (AFAIK) that according to alias this rules, that this conversion should be considered. I mean: Why would "alias this" be OK to get an "int", but doesn't work to get an "S"?

Fixed by: https://github.com/dlang/dmd/pull/8711