Why self:: not $this->. I think code should be uniform.

May be $this->assertInstanceOf will be better?

Accepted $this->, will fix this.

Did not understand about instanceOf. This test explicitly invokes the count() method to be sure, that internals of CookieJar is countable from construction with 7.2

As I understand, you want to check the existence of count method. Am I right? But actually you check that count == 0. So I'm suggesting you to check implementing of the \Countable interface.

No, I wan't to actually invoke the count method. Because PHP 7.2 throws a warning when trying to count a non-countable things like null, which could be done internally (i.e counting internal property which was not initialized) . So I want to check that CookieJar does not throws the warning when calling count directly after constructing.

https://wiki.php.net/rfc/counting_non_countables

The same thing I did with the MockHandler, but we have to fix it (see array default)

You wrote:

So I want to check that CookieJar does not throws the warning when calling count

But actually you

you check that count == 0

So nothing changes in my mind if you will check implementing of the \Countable interface, because it guarantees that calling count won't fail.

But it's kind of a holywar... =)

Having count method implemented does not guarantees that id does not fail. Short example:

class foo implements \Countable {
   private $things;

   public function count() { return count($this->things); }

   public function push($element) { $this->things[] = $element; } 
}

$f = new \foo();
$f->count(); // this will pass on PHP < 7.2 and throw warning on PHP 7.2
$f->push(1);
$f->count(); // this will pass

So having count method implemented does not guarantee that it works properly internally.

Checking that count == 0 is needed because the future phpunit versions will mark tests without assertions as risky, actually I could just call the (new CookieJar())->count() without any assertions.

Can you add a quick not why we are doing this? Just to make sure we find it. Thanks

Done, I've referred your #1823 issue