[RFC] Add PHPStan rules for forbidden methods and unnecessary named arguments

[chr-hertel]

Q	A
Bug fix?	no
New feature?	no
Docs?	no
Issues
License	MIT

I wanted to catch some stuff that a) happens to me and b) could be a rule to enforce while working with named arguments if you ask me.

• Add ForbiddenMethodCallsRule to prevent debugging functions like dump(), var_dump(), print_r(), dd() in production code
• Add ForbidUnnecessaryNamedArgumentsRule to detect unnecessary named arguments when all parameters are provided in correct order
• Support function calls, method calls, and constructor calls
• Handle mixed positional and named arguments
• Register both rules in extension.neon for automatic enforcement across all components

🤖 Generated with Claude Code

[derrabus]

You could consider using https://github.com/spaze/phpstan-disallowed-calls for this.

[OskarStark]

Should be removed

[OskarStark]

Sounds like a nice one for phpstan official repo 🙂

[chr-hertel]

i was rather thinking of PHP CS Fixer - feels a bit too opinionated and more about style than types

[chr-hertel]

Thanks @OskarStark & @derrabus for having a look! Would you in general agree on that rule about named arguments or is it too opinionated?

[derrabus]

Would you in general agree on that rule about named arguments or is it too opinionated?

What does the rule do? Forbid the usage of named arguments if I could've made the same call with ordered arguments?

It is opinionated all right. Not sure what you're trying to solve. If I have a value object like this:

final readonly class Foo {
    public function __construct(
        public string $a,
        public string $b,
        public string $c,
    ) {}
}

Then I wouldn't be allowed to do this?

$foo = new Foo(
    a: 'some value',
    b: 'some value',
    c: 'some value',
);

Meh.

[chr-hertel]

Not sure what you're trying to solve.

To be honest, I just don't want to think about it while reviewing code 😬
(edit: maybe i'm too lazy here :D)

Then I wouldn't be allowed to do this?

$foo = new Foo(
    a: 'some value',
    b: 'some value',
    c: 'some value',
);

Yup, and it's not that I care to much, but was thinking that's the easier rule.

But if we don't mind in general, I can just drop thinking about it while reviewing 😆

[chr-hertel]

no real need, let's close it