Who can explain this deceptively simple destructuring assignment?

Code

var a
[a] = [2]
console.log("a->",a)

var b = {}
[b] = [3]
console.log("b->",b)

Output

a-> 2
b-> [ 3 ]

Why is the second output different? How is the assignment working?

Thanks

Comments

[stereobooster]

You miss semicolon

var b = {};
[b] = [3]
console.log("b->",b)

this would work as expected. It is easy to discover, if you use let:

let b = {}
[b] = [3]
console.log("b->",b)
// Uncaught ReferenceError: can't access lexical declaration 'b' before initialization

Original code:

var b = {}
[b] = [3]

which is the same as

var b = {}[b] = [3]

which is the same as

var b = [3]

[CABrouwers]

Thanks a lot. It seemed absurd to me. All makes sense now!

[myleftshoe]

Wow, nice spot, and worrying for me because I never use semicolons anymore. But I also almost never use var. I guess I've been luck so far.

[F. Selçuk Can]

@stereobooster 's explanation about missing semicolon is correct. In JavaScript in order to terminate a block, {}, you need to type a semicolon, ;.

Unexpected behaviour arises in a number of cases like this. I have never seen your case before.

Another example of this is when you return an object literal which spans multiple lines from a function and do not terminate the return statement after the closing brace of the object literal you return. Interpreter can confuse your object literal with a block since they have same syntax.

Telling that I can say that I have seen that case before :).

My suggestion is to use semicolon after at least object literals. You can choose to use semicolons to terminate only blocks and object literals or even after every line to stay safe.

[James Thomson]

Personally I always use semicolons, I prefer to know exactly where my blocks are terminating. I think it's also that I've been writing JS for so long that I find it unnerving to see no semicolons.