[RFC] Static class

Having read this thread, and the previous one from half a year ago,
I will do so too. In short, we shouldn't be encouraging static
classes as a bag of static functions, that ought to be just
namespaced functions.

Which brings us back to the age-old problem that functions can’t be
autoloaded. Me, I want first-class modules, but until we have those, I
have to settle for classes in the meantime. Scala/Kotlin-like
“companion objects” might be a good all-round substitute though.

There has been some work done on function autoloading too:
https://wiki.php.net/rfc/core-autoloading

cheers,
Derick

My bad, I thought "top post" meant the same thing in mailing lists as it
does in forums. I am now aware of my mistake and it won't happen again. Can
we address my actual points now?

Lanre

There has been some work done on function autoloading too:
https://wiki.php.net/rfc/core-autoloading

Fantastic! I still want modules and/or companion objects of course... I wonder whether the autoloading machinery could be bent to such a purpose, or am I smoking the bad stuff again?

—c

That isn't a top post, look again.And I am not referring to
packages,rather to the ability to define functions that aren't exposed to
the users of a library. C doesn't have packages either yet this is possible
in it. And marking it as internal is not the same thing at all. I am not
going to argue about why a developer would use static methods and whether
that is actually a sign that the developer doesn't understand OOP. But I do
want to get this straight. Instead of actually hiding the function, your
idea is to mark it with @internal (via PHPDoc)? To you, that solves the
problem?

On Wed, Jun 26, 2024 at 1:46 PM Larry Garfield [email protected]
wrote:

In JavaScript/Typescript, we can declare functions that are local to
the current file by not exporting them. In C/C++ this is achieved by
simply writing the function in the implementation file as opposed to
the header. However, achieving a similar level of privacy in PHP
requires using private static methods, or else you incur the overhead
of creating an object for stateless functions. How do you propose
handling such cases using namespaced functions? Can you think of any
scenarios where hiding non-API functions might be necessary? I'm
curious why you keep returning to namespaced functions when they don't
fulfill the same role.

Cheers,
Lanre

Please don't top-post.

PHP has no concept of packages at present, and thus no concept of
package-level scope. A private static method is private not to a file or
package, but to that class. (The class and file usually correlate in
practice, but there's nothing in the language that mandates that.) I would
love to have a concept of packages that we could use for scope, as most
more recent languages have concluded that is the correct level at which to
enforce visibility, NOT objects/classes. That's an entirely separate
question, however.

For marking a function as "package private", the same way you would for a
package-private class today: @internal docblock. It's not ideal (the ideal
would require packages), but it still communicates the intent, has worked
for classes for 15 years at least, and signals "if you use this
class/function/thing and it breaks later, that's on you, don't come crying
to me."

(ab)using static classes to emulate a pseudo-single-class-package is...
abusing static.

To be blunt, 90% of uses of static methods in PHP are a sign the developer
doesn't actually understand OOP and is just writing procedural code with
more steps. The other 10% are either named constructors or cases that can
now be replaced with attributes.

--Larry Garfield

Also I specifically said "we can declare functions that are local to the
current file by not exporting them.", Where did you get "package private"
from?

On Wed, Jun 26, 2024 at 1:46 PM Larry Garfield [email protected]
wrote:

In JavaScript/Typescript, we can declare functions that are local to
the current file by not exporting them. In C/C++ this is achieved by
simply writing the function in the implementation file as opposed to
the header. However, achieving a similar level of privacy in PHP
requires using private static methods, or else you incur the overhead
of creating an object for stateless functions. How do you propose
handling such cases using namespaced functions? Can you think of any
scenarios where hiding non-API functions might be necessary? I'm
curious why you keep returning to namespaced functions when they don't
fulfill the same role.

Cheers,
Lanre

Please don't top-post.

PHP has no concept of packages at present, and thus no concept of
package-level scope. A private static method is private not to a file or
package, but to that class. (The class and file usually correlate in
practice, but there's nothing in the language that mandates that.) I would
love to have a concept of packages that we could use for scope, as most
more recent languages have concluded that is the correct level at which to
enforce visibility, NOT objects/classes. That's an entirely separate
question, however.

For marking a function as "package private", the same way you would for a
package-private class today: @internal docblock. It's not ideal (the ideal
would require packages), but it still communicates the intent, has worked
for classes for 15 years at least, and signals "if you use this
class/function/thing and it breaks later, that's on you, don't come crying
to me."

(ab)using static classes to emulate a pseudo-single-class-package is...
abusing static.

To be blunt, 90% of uses of static methods in PHP are a sign the developer
doesn't actually understand OOP and is just writing procedural code with
more steps. The other 10% are either named constructors or cases that can
now be replaced with attributes.

--Larry Garfield

Hi Larry,

PHP has no concept of packages at present, and thus no concept of package-level scope.

Not entirely true.

PHP has static classes, which are stand-ins for packages for as long as PHP does not inherently support the concept of packages.

Arguing "that is abusing static" without offering an alternative is just Puritanistic logic that ignores pragmatic real world needs.

For marking a function as "package private", the same way you would for a package-private class today: @internal docblock. It's not ideal (the ideal would require packages), but it still communicates the intent, has worked for classes for 15 years at least, and signals "if you use this class/function/thing and it breaks later, that's on you, don't come crying to me."

Not only is @internal docblock not ideal, it sets up those internal properties to be infected with usage as explained by Hyrum's Law[1]. Thus providing @internal docblock as if it were a viable alternative to package private is essentially a non-serious argument.

PHP has no concept of packages at present, and thus no concept of package-level scope. A private static method is private not to a file or package, but to that class. (The class and file usually correlate in practice, but there's nothing in the language that mandates that.) I would love to have a concept of packages that we could use for scope, as most more recent languages have concluded that is the correct level at which to enforce visibility, NOT objects/classes. That's an entirely separate question, however.

If you really want people to stop "abusing static" then drive to address that entirely separate question. Otherwise let others who are more pragmatic give the language features that they need.

As a quick aside, I would argue that PHP should look at its biggest warts — which to me include but are not limited to autoloading and namespaces — and deprecate them to be replaced with a proper package system that leverages a symbol compiler instead of autoloading and that ejects the god-awful choice of using the escape character for package name separation.

But I digress.

To be blunt, 90% of uses of static methods in PHP are a sign the developer doesn't actually understand OOP and is just writing procedural code with more steps. The other 10% are either named constructors or cases that can now be replaced with attributes.

Speaking of best practices, there is a growing contingent of software developers who feel that OOP should not be the future of software[2], e.g. that it was "a good idea at the time" but now no longer.

My point here is that PHP should not let entrenched dogma drive language design but instead pragmatically evolve as other languages are also evolving.

-Mike

[1] https://www.laws-of-software.com/laws/hyrum/ https://www.laws-of-software.com/laws/hyrum/
[2] https://news.ycombinator.com/item?id=8676872 <https://news.ycombinator.com/item?id=8676872

and inheritance is not meant for code reuse.

Just because code reuse in inheritance can be problematic it does not
have to be in all-cases. Moderation in all things. I used that approach for
10+ years and never once had any of the problems people claim about using
inheritance for code reuse. This was likely because my needs were
constrained by the use-case and by nature did not grow out of control with
complexity.

My experience is the opposite. There are subtle bugs I keep running into
with static function and properties causing unexpected behavior. I'm not
against having static classes open by default for the sake of consistency,
though my preference would be to avoid the headache altogether and just
make them final by default so I won't ever have to deal with it. Would
traits not solve the problem of horizontal reuse?

Sounds like a 'you' problem. Can you go into more details about said subtle
bugs? With examples if possible?

Lanre

and inheritance is not meant for code reuse.

Just because code reuse in inheritance can be problematic it does not have to be in all-cases. Moderation in all things. I used that approach for 10+ years and never once had any of the problems people claim about using inheritance for code reuse. This was likely because my needs were constrained by the use-case and by nature did not grow out of control with complexity.

My experience is the opposite. There are subtle bugs I keep running into with static function and properties causing unexpected behavior. I'm not against having static classes open by default for the sake of consistency, though my preference would be to avoid the headache altogether and just make them final by default so I won't ever have to deal with it. Would traits not solve the problem of horizontal reuse?

Hi Lynn,

I understand it's frustrating when there are bugs/unexpected behaviour, but wouldn't the (existing) ability to add the final keyword to a class solve those issues for you, without preventing others from using the capabilities of parent/child class relationships?

Cheers

Stephen

Hi,

Hi guys,

Given a primary purpose for being able to declare a class static is to signal intent, disallowing abstract static classes seems at odds with that goal.
What is the intent of abstract static? How is such intent different from just static?

Sorry I should have been clearer. It isn't that abstract represents an intent, it's that adding static is mostly about conveying intent - and thus there's no technical reason that any existing class with only static members can't be marked as a static class, so why introduce an artificial barrier if that class happens to already be abstract?

It's already possible to have a class with only static members, that is abstract, and call the public static methods of it. It's also possible to have a class that partially implements an interface that has static methods on it. Unless your RFC is going to propose preventing calling static methods on all abstract classes it's a needless complication when there's no technical reason to do so.

I agree that the static keyword is a much better fit here, however there is one other aspect here that may come into it (as much as I prefer the keyword approach): the Attribute approach is backwards compatible (with a polyfill) to let code using this feature also run on previous PHP releases. Given that this is mostly intended as a way to signal intent about a pattern that's already in use, this may be significant for library authors.

Personally (as a library author) I don't think that ability is worth the weirdness of an attribute vs a keyword, but it may be more important for others who are voting, and I'd rather have the feature with slightly odd syntax rather than not at all.
When I first saw the proposal to use an attribute instead of the keyword, I thought it absurd, but this idea has now been entertained by three people, and for the first time I have seen a compelling argument in favour of (early adoption). I must admit, I was too quick to judge this one. I had not considered that libraries will still not be able to use the static modifier with classes unless and until they drop support for PHP < 8.4, which may take a while! Of course, it is still of real benefit to first-party proprietary projects whom have upgraded. Nevertheless, the allure of early adoption is curious, and made me wonder whether we could do both, just to support early adoption in a backwards-compatible manner. However, this would be unprecedented and most likely not accepted; never before has the same feature been implemented two ways just to appease early adopters. I think the best compromise would be, for anyone so eager, to implement such support in community tools, e.g. PHPStan. That is, it should be perfectly possible to enforce the core semantics of the static class feature with a userland attribute and the necessary logic in static analysis tools, provided the library is well behaved and doesn't do anything too weird with variable variables, references, reflection or unserialize() to deliberately circumvent the restriction.

Yes it should be possible to use a user land attribute or phpdoc tag or what have, but that's already true. You've essentially just made a case against your own RFC - that it can mostly be done already in userland.

The entire appeal of a builtin attribute would be that it's something you could apply now (as in today, June 2024) and know that it won't break anything, while having some {compile,run} time benefits in later versions (e.g. enforcing static-only, and inability to instantiate).

Like I said, I much prefer the keyword syntax - but I also recognise that others may have different priorities in terms of supporting older language versions, which is why I think that aspect is worth consideration (perhaps a secondary vote, or an informal vote to gauge consensus?), because too many good RFCs get rejected over small details.

Cheers,
Bilge

Cheers

Stephen

Nevertheless, the allure of early adoption is curious, and made me wonder whether we could do both, just to support early adoption in a backwards-compatible manner.

And that is probably the best solution I've heard regarding this quandary.

It could potentially apply with other keywords in the future as well.

#fwiw

-Mike

• Why is it a class-level flag and not an attribute (similar to the
  #[Override] attribute in PHP 8.3) ?

From my perspective that would create much confusion among every PHP developer who has not committed to memory when to use static as a class keyword and when to use it as an attribute.

Given the concept already exists in keywords I would strongly argue that it makes no sense to introduce as an attributes in core PHP.

Attributes are great for concepts new to PHP, IMO, but not for existing concepts already part of PHP implemented with keywords.

• The main purpose of the abstract keyword is to prevent a class to be instantiated, which (in case of static class) is more semantically described by the static marker. Beyond that, it just allows to declare a method that, if implemented by a subclass, should have a compatible signature. Most notably, it does not prevent the other static members of the class to be used directly.

Given a primary purpose for being able to declare a class static is to signal intent, disallowing abstract static classes seems at odds with that goal.

Of course it is currently true that static methods of abstract classes can be called from outside a class and its class hierarchy, so if we allow declaring abstract static classes then it would never in the future be possible to lock down calls of static methods to those abstract classes.

So IMO it would be better to disallow calling static methods from outside a declared abstract static class and its inheritance hierarchy as part of this RFC. That would be backward compatible since there are currently no classes that are declared in that manner. Doing otherwise would force those who want to declare a class as both static and abstract to have to make a choice rather than being able to signal their full intent. Which brings us back to the "implied" vs. "explicitly declared" bifurcation I mentioned in a prior email.

BTW, I am assuming it is technically possible to disallow calling methods for classes declared both abstract and static without considerable difficulty in implementation and without creating a significant drain on performance.

-Mike

P.S. I would argue the same for readonly static properties, but as that seems those would require an involved discussion about the exact rules to implement I am demurring on that topic.

Hi Mike,

So IMO it would be better to disallow calling static methods from outside a declared abstract static class

Can you explain what you mean by this, or more specifically the "why". What (implied wrong/bad) scenario are you trying to prevent, by disallowing a static method on an abstract class to be called?

As you point out, it's already possible to call (public) static methods on abstract classes, the same as it's possible to call (public) static methods without instantiating a regular class.

So why should static abstract classes be different? If the static method shouldn't be called publicly, it shouldn't be public. Are you suggesting that public static methods shouldn't be callable on any abstract class? If so, this sounds like a huge BC break, and if not, it sounds like a confusing "if-then-but-why" scenario.

Thank you for the question.

I was not specifically advocating static methods to be disallowed to be called on an abstract class. I was instead addressing the desire by another list member to have classes declared static not also be able to be declared abstract. That person made the argument that if 'abstract static` did not actually have any effect then it should not be allowed by the RFC.

What affect are we talking about? If it is both abstract and static then it would seem to me the only effect such a declaration could have would be to disallow the calling of static methods using the named abstract static class, e.g.:

abstract static class Base {
static public foo() {
echo "foo() called\n"
}
}
static class Concrete extends Base {}

Concrete::foo(); // works
Base::foo(); // fails

My only real argument for disallowing static method calls on abstract static classes is that if we didn't do so initially we would never be able to disallow because of BC.

In summary, my argument was really only about ensuring we allow "abstract static" declaration so developers can signal intent and also to address the objections of the other list member. But if given the option, I think it would be good to disallow calling static methods on abstract static classes if only to reserve the right to allow in the future vs. closing that door immediately. Still, that latter is not the hill for me to die on.

I agree that the static keyword is a much better fit here, however there is one other aspect here that may come into it (as much as I prefer the keyword approach): the Attribute approach is backwards compatible (with a polyfill) to let code using this feature also run on previous PHP releases. Given that this is mostly intended as a way to signal intent about a pattern that's already in use, this may be significant for library authors.

Well, as previously stated, I think that would be confusing.

Though I do see your reasoning for wanting it to be an attribute. However...

Personally (as a library author) I don't think that ability is worth the weirdness of an attribute vs a keyword, but it may be more important for others who are voting, and I'd rather have the feature with slightly odd syntax rather than not at all.

I too am not convinced that the ongoing confusion would be better than the temporary need to support older versions of PHP with the same source file.

Given that a build step could address the syntax there are certainly other new language features that require developers to bifurcate their libraries to support different versions, why make this one unique?

That said, it could be a voting option for the RFC?

-Mike

Hi Mike,

This is an example of code that works today (and all the way back to 5.0): https://3v4l.org/4EKo2

The class hierarchy embody the type of classes this RFC is about: only static members, no instantiation.

The implemented methods can be called statically, regardless of whether the class they're implemented in is abstract or not. The abstract methods cannot be called directly.

So these classes would be a candidate for the static class keyword (or Attribute) - except they can't, if calls to implemented methods on abstract classes are disallowed. Because the Base::a() method has been publicly callable, for potentially as long as <checks notes> 20 years next month.

My point here is that if someone wants to prohibit calling public static methods on abstract classes with the static keyword, that's going be inconsistent with how it's worked for the last 20 years (i.e. on classes that were 'static' in intent but not syntactically), or if it applies the change everywhere it's going to be a BC break.

Re: attribute vs keyword,

Like I said - I'm not personally in favour of the attribute approach; I'm simply making the point (playing devils advocate if you will) that by the very nature of how attributes work, they're more friendly to supporting older versions of the language with a single code base. If the general consensus is that it isn't a concern, great. But this wouldn't be the first time we've seen a relatively minor syntax choice derail an otherwise useful RFC.

Cheers

Stephen

Hi Tim,

and I'd rather have the feature with slightly odd syntax rather than
not at all.

Absolutely not. PHP is already odd enough as it is. We should not add
additional oddness in the interest of pushing a new screw-shaped
feature in with a sledgehammer. Any additional oddity makes it harder
for Junior developers and developers coming from a different language
to learn PHP and also required the developer to learn the language
details by heart instead of being able to rely on their intuition.
Don't worry, I couldn't bring myself to seriously entertain using an
attribute instead of a keyword. I'll update the RFC to include it as a
rejected feature.
I'm already leaning towards "Don't want static classes"

Why? It is important to understand what your misgiving are in case there
anything we can do to assuage them.

Cheers,
Bilge

On Mon, Jul 8, 2024 at 1:12 PM Stephen Reay [email protected]
wrote:

So really the question should be: why do you feel the need to disallow a
combination that has no technical issues, and works already? If you're
going to disallow abstract why not disallow any class inheritance at all?
Because that's what abstract relates to: class hierarchy. Why remove some
support but keep the rest?

Exactly what I wanted to say for this topic. Have a look at this example
that could be very similar in production code in some systems:

<?php

interface A {
public static function run();
public static function stop();
}

abstract class B implements A {
public static function run() {
static::create();
static::start();
}
}

class C extends B {
public static function create() {
echo "Create\n";
}

public static function start() {
    echo "Start\n";
}

public static function stop() {
    echo "Stop\n";
}

}

function execute($class) {
if (!is_subclass_of($class, A::class)) {
throw new RuntimeException("Class must implement A interface");
}

$class::run();
$class::stop();

};

execute(C::class);

If we support inheritance for static classes, we should allow static on
both interface and abstract classes.

Alex