I had to alter the ignoreErrors entries for the Mixin trait, to accomodate this change. It doesn't get PHPStan issues to 0 - but does get them back to where they were before this PR.

I needed to swap:

'#^Undefined variable: \$this$#'
'#^Variable \$this in isset\(\) is never defined\.$#'

for

'#^Variable \$this in isset\(\) always exists and is not nullable\.$#'

This is because this (Macro|CarbonInterval)::macro method now believes $this is static. In reality this isn't actually true: given the rather unsual context in which this sits, $this could actually be static or undefined (for reference, it previously would have been though of as just undefined). I still believe this is the best path forward (despite having to ignore an error), as the situation in which $this is undefined is not exposed to userland code - and this change does resolve my original issue.

I think to resolve this properly (as in, without needing to ignore an error) - the logic that handles "whether we're in a static context" (see below) would need to rejigged. I think this would be out of the scope of this PR?

private static function loadMixinTrait(string $trait): void
{
    $context = eval(self::getAnonymousClassCodeForTrait($trait));
    $className = \get_class($context);
    $baseClass = static::class;

    foreach (self::getMixableMethods($context) as $name) {
        $closureBase = Closure::fromCallable([$context, $name]);

        static::macro($name, function (...$parameters) use ($closureBase, $className, $baseClass) {
            // This is the "whether we're in a static context" I'm referring to
            $downContext = isset($this) ? ($this) : new $baseClass();
            $context = isset($this) ? $this->cast($className) : new $className();
    // ...

I only added @param-closure-this to the methods that were required to satisfy PHPStan. In reality I think this should also be added to:

1. src/Carbon/Factory::macro
2. src/Carbon/CarbonInterface::macro
3. src/Carbon/CarbonPeriod::macro

However I've left them out to keep the PR as slim as possible (feels slight pointless after writing this wall of text 😅) - happy to be directed otherwise.

Thanks!

Failures all look unrelated to this PR?

Only the error ignore is to be removed from the phpstan config, the rest is fine.

I think the Laravel test failures are due to also unrelated? see this.

Edit: sorry I misread that - I'll look into the PHPStan error.

I've resolved that PHPStan issue. It occurs (for me) on master (with none of my changes applied) - which is why I didn't do anything about it initially.

I can't immediately see how it could be related to my change?