diff options
Diffstat (limited to 'src/Sema.zig')
| -rw-r--r-- | src/Sema.zig | 16 |
1 files changed, 9 insertions, 7 deletions
diff --git a/src/Sema.zig b/src/Sema.zig index 13a1c1cb3d..866c242f15 100644 --- a/src/Sema.zig +++ b/src/Sema.zig @@ -66,11 +66,14 @@ comptime_args_fn_inst: Zir.Inst.Index = 0, /// extra hash table lookup in the `monomorphed_funcs` set. /// Sema will set this to null when it takes ownership. preallocated_new_func: ?*Module.Fn = null, -/// The key is `constant` AIR instructions to types that must be fully resolved -/// after the current function body analysis is done. -/// TODO: after upgrading to use InternPool change the key here to be an -/// InternPool value index. -types_to_resolve: std.ArrayListUnmanaged(Air.Inst.Ref) = .{}, +/// The key is types that must be fully resolved prior to machine code +/// generation pass. Types are added to this set when resolving them +/// immediately could cause a dependency loop, but they do need to be resolved +/// before machine code generation passes process the AIR. +/// It would work fine if this were an array list instead of an array hash map. +/// I chose array hash map with the intention to save time by omitting +/// duplicates. +types_to_resolve: std.AutoArrayHashMapUnmanaged(InternPool.Index, void) = .{}, /// These are lazily created runtime blocks from block_inline instructions. /// They are created when an break_inline passes through a runtime condition, because /// Sema must convert comptime control flow to runtime control flow, which means @@ -34085,8 +34088,7 @@ fn anonStructFieldIndex( } fn queueFullTypeResolution(sema: *Sema, ty: Type) !void { - const inst_ref = try sema.addType(ty); - try sema.types_to_resolve.append(sema.gpa, inst_ref); + try sema.types_to_resolve.put(sema.gpa, ty.toIntern(), {}); } fn intAdd(sema: *Sema, lhs: Value, rhs: Value, ty: Type) !Value { |