Add an example for closure with an empty body

[mspirkov]

@keradus suggested clarifying the situation with closures that have an empty body: PHP-CS-Fixer/PHP-CS-Fixer#9329 (comment)

I agree with him. I think the Closures section lacks an example with an empty body. It seems logical to add an example similar to the one in the Classes, Properties, and Methods section:

per-coding-style/spec.md

Lines 592 to 603 in c83ab63

	If a function or method contains no statements or comments (such as an empty no-op implementation or when using
	constructor property promotion), then the body SHOULD be abbreviated as `{}` and placed on the same
	line as the previous symbol, separated by a space. For example:
	```php
	class Point
	{
	public function __construct(private int $x, private int $y) {}
	// ...
	}
	```

What do you think about this?

[keradus]

would recommend MUST
(but happy with both)

[mspirkov]

I think it should be similar to the Classes, Properties, and Methods section:

per-coding-style/spec.md

Lines 592 to 593 in c83ab63

	If a function or method contains no statements or comments (such as an empty no-op implementation or when using
	constructor property promotion), then the body SHOULD be abbreviated as `{}` and placed on the same

[keradus]

would recommend MUST in both places then ;)

[keradus]

would love this to be explicitly clarified ! 🎉

[KorvinSzanto]

This isn't so much a clarification as it is a change to requirements. Closures currently have the following requirement:

The opening brace MUST go on the same line, and the closing brace MUST go on
the next line following the body.

And abbreviated {} style is explicitly limited to classes, methods, and functions: #44

I think it'd be okay to make this change, but it's important to be clear that it's not just a clarification.

That said, when would someone actually want to do this instead of using a short function which already allows single line declaration?

IE:

$noOpClosure = fn() => null;

[jrfnl]

This isn't so much a clarification as it is a change to requirements.

Agreed, so this would need a new major release.

[mspirkov]

This isn't so much a clarification as it is a change to requirements.

From the outside, it seems that this case was simply forgotten to be described, since it is written explicitly for class methods and functions. But I agree with you - this is actually a new requirement.

[mspirkov]

That said, when would someone actually want to do this instead of using a short function which already allows single line declaration?

I think this is a real case, because:

1. In PHP versions lower than 7.4, we cannot use fn
2. I think it's a matter of taste. I mean using the fn() => null or function () {}

[Crell]

For context, the reason we went with a SHOULD for classes was to avoid it being a hard BC break. It allowed existing code to continue to be legal, while formatters changed their defaults gradually to single-line it.

Would a SHOULD here be safe enough to only trigger a 3.1, not a 4.0, since it leaves existing code still legal? I'd prefer to not tag a 4.0 if possible, though I agree the same logic should apply to anything with a body.

That said, @KorvinSzanto is correct that the use case in practice seems weird, when short closures are even nicer and have no such concerns. To @mspirkov's point, 7.4 is long-since abandoned so we don't really care about its limitations at this point. 😄

[keradus]

so this would need a new major release

I was seeing this as forgotten case, as examples were not mentioning it.
If that would be concluded to be bc breaker, please let's documment the status quo for empty body for closures, it would be already better than no explicit description of such case.

and potential breaking change could go to new major, separately, if aligned

[keradus]

when would someone actually want to do this instead of using a short function which already allows single line declaration?

this was actually coming as request to PHP CS Fixer from PHP community.
ie, 264 usages in Symfony v8 (git grep function | grep ") {}" | wc -l)

If usage of given syntax is allowed, it's great to document style for it explicitly. If you believe it's better to recommend avoiding using function () {} and recommend to use fn() => null instead, totally valid - yet same here, please let's have it documented ;)

[Crell]

Let's go with "if a closure contains no statements, MUST use the short-hand syntax" (with an example). That neatly sidesteps the BC issue on braces, and gives a clear standard of behavior. Also, we already have a clause somewhere that says "ignore things that don't apply to your PHP version", so anyone who cares about <7.4 for some reason has nothing to do.

[mspirkov]

Do I understand correctly that we are talking about fn() => null?

[Crell]

Yes, that.

[keradus]

@Crell , if I understand idea of MUST vs SHALL in BC break context for PER-CS, I believe your suggestion is same BC as defining spaces/newline for function () {}
so far, function () {\n} is valid according to PER-CS, and your suggestion would make it invalid - so also BC break, no?

it is not that its reserved for PHP <7.4 only, as function () {} is valid syntax on PHP 8.5 and can be used as is, no?

[Crell]

Let's go with "if a closure contains no statements, MUST use the short-hand syntax" (with an example). That neatly sidesteps the BC issue on braces, and gives a clear standard of behavior. Also, we already have a clause somewhere that says "ignore things that don't apply to your PHP version", so anyone who cares about <7.4 for some reason has nothing to do.

By shorthand, we mean fn() => null.

[jrfnl]

By shorthand, we mean fn() => null.

May I suggest making that explicit in the text ? I.e. "if a closure contains no statements, it MUST be declared as an arrow function"

[mspirkov]

@Crell Please take a look

[Crell]

I will merge this in a few days if no members of the WG speak up in opposition.