Imperative control flow
There are two key categories of control flow:
-
declarative, when the structure of some value guides control and the selection of the next expression to evaluate, like in
if
andswitch
expressions; -
imperative where control changes abruptly according to a programmer’s command, abondoning regular control flow; examples are
break
andcontinue
, but alsoreturn
andthrow
.
Imperative control flow often goes hand-in-hand with state changes and other flavors of side-effects, such as error handling and input/output.
Early return
from func
Normally, the result of a function is the value of its body. Sometimes, during evaluation of the body, the result is available before the end of evaluation. In such situations the return ⟨exp⟩
construct can be used to abandon the rest of the computation and immediately exit the function with a result.
Similarly, where permitted, throw
may be used to abandon a computation with an error.
When a function has unit result type, the shorthand return
may be used instead of the equivalent return ()
.
Loops and labels
Motoko provides several kinds of repetition constructs, including:
-
for
expressions for iterating over members of structured data. -
loop
expressions for programmatic repetition (optionally with termination condition). -
while
loops for programmatic repetition with entry condition.
Any of these can be prefixed with a label ⟨name⟩
qualifier to give the loop a symbolic name. Named loops are useful for imperatively changing control flow to continue from the entry or exit of the named loop.
-
re-entering the loop with
continue ⟨name⟩
, or -
exiting the loop altogether with
break ⟨name⟩
.
In the following example, the for
expression loops over characters of some text and abandons iteration as soon as an exclamation sign is encountered.
import Debug "mo:base/Debug";
label letters for (c in "ran!!dom".chars()) {
Debug.print(debug_show(c));
if (c == '!') { break letters };
// ...
}
Labeled expressions
There are two other facets to label
s that are less mainstream, but come in handy in certain situations:
-
label
s can be typed -
any expression (not just loops) can be named by prefixing it with a label;
break
allows one to short-circuit the expression’s evaluation by providing an immediate value for its result. (This is similar to exiting a function early usingreturn
, but without the overhead of declaring and calling a function.)
The syntax for type-annotated labels is label ⟨name⟩ : ⟨type⟩ ⟨expr⟩
, signifying that any expression can be exited using a break ⟨name⟩ ⟨alt-expr⟩
construct that returns the value of <alt-expr>
as the value of ⟨expr⟩
, short-circuiting evaluation of <expr>
.
Judicious use of these constructs allows the programmer to focus on the primary program logic and handle exceptional case via break
import Text "mo:base/Text";
import Iter "mo:base/Iter";
type Host = Text;
let formInput = "us@dfn";
let address = label exit : ?(Text, Host) {
let splitted = Text.split(formInput, #char '@');
let array = Iter.toArray<Text>(splitted);
if (array.size() != 2) { break exit(null) };
let account = array[0];
let host = array[1];
// if (not (parseHost(host))) { break exit(null) };
?(account, host)
}
Naturally, labeled common expressions don’t allow continue
. In terms of typing, both ⟨expr⟩
and ⟨alt-expr⟩
's types must conform with the label’s declared ⟨type⟩
. If a label is only given a ⟨name⟩
, then its ⟨type⟩
defaults to unit (()
). Similarly a break
without an ⟨alt-expr⟩
is shorthand for the value unit (()
).
Option blocks and null breaks
Like many other high-level languages, Motoko lets you opt in to null
values, tracking possible occurences of null
values using option types of the form ?T
.
This is to both to encourage you to avoid using null
values when possible, and to consider the possiblity of null
values when necessary.
The latter could be cumbersome, if the only way to test a value for null
were with a verbose switch
expression, but
Motoko simplifies the handling of option types with some dedicated syntax: option blocks and null breaks.
The option block, do ? <block>
, produces a value of type ?T
, when block <block>
has type T
and, importantly, introduces the possibility of a break from <block>
.
Within a do ? <block>
, the null break <exp> !
, tests whether the result of the expression, '<exp>', of unrelated option type, ?U
, is null
.
If the result <exp>
is null
, control immediately exits the do ? <block>
with value null
.
Otherwise, the result of <exp>
must be an option value ?v
, and evaluation of <exp> !
proceeds with its contents, v
(of type U
).
As realistic example, we give the definition of a simple function `eval`uating numeric `Exp`ressions built from natural numbers, division and a zero test, encoded as a variant type:
type Exp = {
#Lit : Nat;
#Div : (Exp, Exp);
#IfZero : (Exp, Exp, Exp);
};
func eval(e : Exp) : ? Nat {
do ? {
switch e {
case (#Lit n) { n };
case (#Div (e1, e2)) {
let v1 = eval e1 !;
let v2 = eval e2 !;
if (v2 == 0)
null !
else v1 / v2
};
case (#IfZero (e1, e2, e3)) {
if (eval e1 ! == 0)
eval e2 !
else
eval e3 !
};
};
};
}
To guard against division by 0
without trapping, the eval
function returns an option result, using null
to indicate failure.
Each recursive call is checked for null
using !
, immediately exiting the outer do ? block
, and thus the function itself, with null
, when a result is null
.
(As an exercise that illustrates the concision of option blocks, you might want to try rewriting eval
using a labeled expression and explicit switches for each null break.)
Repetition with loop
The simplest way to indefinitely repeat a sequence of imperative expressions is by using a loop
construct
loop { ⟨expr1⟩; ⟨expr2⟩; ... }
The loop can only be abandoned with a return
or break
construct.
A re-entry condition can be affixed to allow a conditional repetition of the loop with loop ⟨body⟩ while ⟨cond⟩
.
The body of such a loop is always executed at least once.
while
loops with precondition
Sometimes an entry condition is needed to guard the first execution of a loop. For this kind of repetition the while ⟨cond⟩ ⟨body⟩
-flavor is available
while (earned < need) { earned += earn() };
Unlike a loop
, the body of a while
loop may never be executed.
for
loops for iteration
An iteration over elements of some homogeneous collection can be performed using a for
loop. The values are drawn from an iterator and bound to the loop pattern in turn.
let carsInStock = [
("Buick", 2020, 23.000),
("Toyota", 2019, 17.500),
("Audi", 2020, 34.900)
];
var inventory : { var value : Float } = { var value = 0.0 };
for ((model, year, price) in carsInStock.vals()) {
inventory.value += price;
};
inventory
Using range
with a for
loop
The range
function produces an iterator (of type Iter<Nat>
) with the given lower and upper bound, inclusive.
The following loop example prints the numbers 0
through 10
over its eleven iterations:
import Iter "mo:base/Iter";
import Debug "mo:base/Debug";
var i = 0;
for (j in Iter.range(0, 10)) {
Debug.print(debug_show(j));
assert(j == i);
i += 1;
};
assert(i == 11);
More generally, the function range
is a class
that constructs iterators over sequences of natural numbers. Each such iterator has type Iter<Nat>
.
As a constructor function, range
has a function type:
(lower : Nat, upper : Int) -> Iter<Nat>
Where Iter<Nat>
is an iterator object type with a next
method that produces optional elements, each of type ?Nat
:
type Iter<A> = {next : () -> ?A};
For each invocation, next
returns an optional element (of type
?Nat
).
The value null
indicates that the iteration sequence has terminated.
Until reaching null
, each non-null
value, of the form ?
n for some number n, contains the next successive element in the iteration sequence.
Using revRange
Like range
, the function revRange
is a class
that constructs iterators (each of type Iter<Int>
).
As a constructor function, it has a function type:
(upper : Int, lower : Int) -> Iter<Int>
Unlike range
, the revRange
function descends in its iteration sequence, from an initial upper bound to a final lower bound.
Using iterators of specific data structures
Many built-in data structures come with pre-defined iterators. Below table lists them
Type | Name | Iterator | Elements | Element type |
---|---|---|---|---|
|
array of |
|
the array’s members |
|
|
array of |
|
the array’s valid indices |
|
|
mutable array of |
|
the array’s members |
|
|
mutable array of |
|
the array’s valid indices |
|
|
text |
|
the text’s characters |
|
|
blob |
|
the blob’s bytes |
|
User-defined data structures can define their own iterators. As long they conform with the Iter<A>
type for some element type A
, these behave like the built-in ones and can be consumed with ordinary for
-loops.