[4.0] Use null coalescing operator where applicable

[SharkyKZ]

Pull Request for Issue # .

Summary of Changes

Use null coalescing operator over ternary operator and instanceof checks in methods with nullable class typehints.

Testing Instructions

1. Code review.
2. Browse around in frontend and backend.

Expected result AFTER applying this Pull Request

2. Joomla still works.

Documentation Changes Required

No.

[richard67]

Fore some reason this PR doesn't appear in the issue tracker? @zero-24 Any idea? Or who could know?

[mbabker]

?? is essentially a shorthand for isset($foo) && !empty($foo) ? $foo : $bar

The $db variable is always set, even if null. So the shorthand ?: is correct here.

Null coalescing really should only be used in cases where it is uncertain if a reference (array key or variable) will be defined.

All of that to say that what’s there pre-patch is correct for these types of uses and while the null coalescing operator might work to me it raises a flag in a code review because then I’m looking for a code path where something on the left side isn’t defined at all.

[SharkyKZ]

?? is shorthand for isset(). It allows anything but null. Both ?? are ?: fine in terms of functionality. The idea here was to use ?? sort of as a strict comparison against null. But maybe just use === null instead. Anyways I'm just nitpicking here.

[mbabker]

The $db = $db instanceof DatabaseDriver ? $db : Factory::getDbo(); stuff absolutely should be shorthanded to $db = $db ?: Factory::getDbo();.

But personally, I would favor ?: over ?? always unless you are dealing with potentially unassigned variables or array keys or things like that.

So in places where there are constructors with a signature of __construct(array $config), those are the perfect places to use ?? to handle things like $this->foo = $config['foo'] ?? $saneDefault;.

I get in all of the cases in this PR that ?? and ?: result in functionally equivalent code, but to people actually reading the code and understanding the operators, whenever I see ?? personally I'm next looking to see why it's shorthanding the left side of that operation needs some form of isset check (and no, I don't consider isset checks as necessary on $foo = null assignments, those variables are actually defined but they don't have values).

[SharkyKZ]

To me ?? looks like a strict comparison against null. But I guess it's a matter of preference. Going to close this.

[zero-24]

Fore some reason this PR doesn't appear in the issue tracker? @zero-24 Any idea? Or who could know?

Please open for such issues an issue here: joomla/jissues. I have no access to the issue tracker to check the reason behind it nor do i want to have any access there ;) The last time it was an issue with webhooks that run on error and it seemed to be an upstream issue from GitHub but best is to open an issue there so this can be checked.