Bug 12261 – [REG2.066a] alias T cannot bind to templated type

struct A1(T){} template isAliasable(alias T) { enum isAliasable=true; } template isAliasable(T) { enum isAliasable=false; } void main(){ alias foo=A1!double; //this works static assert(isAliasable!(A1!double)); //ok in 2.064.2; CT error in 2.065 even though previous line ok }

Actually this also fails with: struct A2{} static assert(isAliasable!(A2)); //ok in 2.064.2; CT error in 2.065 Is that a bugfix or a bug?

This only manifests in git, not the released 2.065, right?

Introduced in https://github.com/D-Programming-Language/dmd/pull/3210. Looking at that pull's testcases, this might be an intentional change - (T) is more specialized than (alias T), as all types can be aliases but not all aliases are types.

(In reply to comment #3) > Introduced in https://github.com/D-Programming-Language/dmd/pull/3210. > > Looking at that pull's testcases, this might be an intentional change - (T) is > more specialized than (alias T), as all types can be aliases but not all > aliases are types. Yes, the current behavior has practical benefit that: template isAliasable(alias T) { pragma(msg, T); } template isAliasable(T) { pragma(msg, T); } struct A1(T) {} struct A2{} alias a1 = isAliasable!(const A1!double); alias a2 = isAliasable!(const A2); With 2.065: A1!double A2 --> type qualifier information has lost With 2.066a: const(A1!double) const(A2) --> accessible to the full information

Marked as invalid.

*** Issue 13296 has been marked as a duplicate of this issue. ***