** and ++ operators force comptime on operands

closes #1707

src/ir.cpp

@@ -3182,8 +3182,13 @@ static IrInstruction *ir_gen_block(IrBuilder *irb, Scope *parent_scope, AstNode
 }
 
 static IrInstruction *ir_gen_bin_op_id(IrBuilder *irb, Scope *scope, AstNode *node, IrBinOp op_id) {
-    IrInstruction *op1 = ir_gen_node(irb, node->data.bin_op_expr.op1, scope);
-    IrInstruction *op2 = ir_gen_node(irb, node->data.bin_op_expr.op2, scope);
+    Scope *inner_scope = scope;
+    if (op_id == IrBinOpArrayCat || op_id == IrBinOpArrayMult) {
+        inner_scope = create_comptime_scope(irb->codegen, node, scope);
+    }
+
+    IrInstruction *op1 = ir_gen_node(irb, node->data.bin_op_expr.op1, inner_scope);
+    IrInstruction *op2 = ir_gen_node(irb, node->data.bin_op_expr.op2, inner_scope);
 
     if (op1 == irb->codegen->invalid_instruction || op2 == irb->codegen->invalid_instruction)
         return irb->codegen->invalid_instruction;

test/cases/eval.zig

@@ -762,3 +762,21 @@ test "*align(1) u16 is the same as *align(1:0:2) u16" {
         //assert(*align(:0:2) u16 == *u16);
     }
 }
+
+test "array concatenation forces comptime" {
+    var a = oneItem(3) ++ oneItem(4);
+    assert(std.mem.eql(i32, a, []i32{3, 4}));
+}
+
+test "array multiplication forces comptime" {
+    var a = oneItem(3) ** scalar(2);
+    assert(std.mem.eql(i32, a, []i32{3, 3}));
+}
+
+fn oneItem(x: i32) [1]i32 {
+    return []i32{x};
+}
+
+fn scalar(x: u32) u32 {
+    return x;
+}