___

[lexidor]

The following issues are stale and keeping the issue count on hhvm high:

Issue reference	Verdict	Reason for closing
#1066	Stale	This refers to the performance of HHVM back in 2013.
#1447	PHPism?	#7926 (unmerged) addressed this issue. Unsure how to feel about modifying a file during a request. This does sound like a PHPism to me.
#1605	Implementation detail of PHP	You should not be relying on the order of constants in which constants are defined in get_defined_constants.
#1589	Famously unmaintained	It looks like the last non-cleanup commit in the code was about 22 months ago. (Jan 2014)
#1669	PHPism	All hack type hints are strictly adhered to.
#2060	Programmer Error	Registering a private function to a callback based api is not the fault of the api, but rather a programmer error.
#2108	Deprecated	From a conversation with a Facebook employee -> "filters are gone, wrappers that can re-enter the VM are gone." Feels like it applies here too.
#2227	Fixed	I can see an HHI for WeakRef, so it exists now.
#2260	Fixed	Tried in nightly and it works there. Assume that this works everywhere else too.
#2268	PHP feature unsupported	You can't use ArrayAccess<_, _> anyhow since hack will tell you that you are using a subscript operator on something that is not a KeyedContainer<_, _> $foo = new Foo(); $foo['a'] = 'b';``Typing[4005] This is not an object of type KeyedContainer, this is an object of exactly the class Foo.
#2406	hhvm-autoload, HH\fact_parse, and hhast exist	It seems like @fredemmott was trying to do what hhvm-autoload is doing now. I presume that hhvm is providing the required functions and hooks to perform this, else how would hhvm-autoload do it?
#2560	Works?	I tried the example from PHP docs on PHP authentication and it worked. It is unclear to me which fields may still be missing. I was able to see most of them using hhvm -m server and vardumping the keys of $_SERVER.
#2692	Fixed?	My own website is serving some static .json files just fine. That might be because of how my op configured it though. It doesn't seem like this is an issue anymore.
#2730	Can be invoked using parens	blog mentions ($this->callable_property)(...$args)
#2926	Untyped hack	require_once (like echo and friends) is an untyped statement. This is expected behavior. You should not be requiring things yourself if you can help it. Autoloading is baked into the core of hhvm. The only thing that really needs to be required these days is /vendor/bin/autoload.hack.
#2957	Sample code unsupported	It is hard to figure out what sql lib was used. It might have been mysqli. Even if it was something else, the given sample code is php that exploits many deprecated features. I don't think that this is something that we can realistically still use.
#2976	Stub	#2976 (comment)
#3015	HHI change	call_user_func_array takes Container<_>, stdClass is stopped at typecheck time.
#3193	PHPism	We are strict about the difference between static methods and instance methods.
#3273	Updated	This describes the current behavior of // strict.

Okay, this took waaay to long already. Will flag a couple of issues every time I create a new one. I want the issue count to slowly go down to something that is more manageable.

[ecreeth]

Issue	Reason
#6954	PHP related
#6913	PHP 5 related
#6183	PHP 5 related
#6026	PHP 5 related
#5786	PHP 5 related
#5541	PHP 5 related
#3874...?	PHP 5 related

[lexidor]

@ecreeth Could you explain your rationale for #3874? Is this okay to close because this behavior is also correct, but just different from PHP? Or are we merely closing this because of the PHP comparison?

(new ReflectionFunction($closure))->isClosure();

The issue says that this returns false. Does it still do that today? That seems like something worth fixing.

[ecreeth]

@lexidor This is due to the comparison with php and the year it was created

[lexidor]

The quirks that were mentioned in the issue are still accurate to this day. The example does not work out of the box, since ReflectionFunction::getDefiningFunction() is not a thing, you need ReflectionFunction::getDeclaringFunction().

HHVM + Hack 4.38.0-dev (rel)

bool(false)
NULL
object(ReflectionClass) (2) {
  ["name"]=>
  string(27) "Closure$Ship::getClosure;10"
  ["obj":"ReflectionClass":private]=>
  NULL
}

PHP + php 7.4.1 (cli)

bool(true)
object(ReflectionClass)#2 (1) {
  ["name"]=>
  string(4) "Ship"
}
object(ReflectionClass)#2 (1) {
  ["name"]=>
  string(4) "Ship"
}

I don't speak Reflectionista and I don't know if anyone uses this / cares.
It is worth noting that ReflectionFunction::getClosureScopeClass() does not exist in the hhi's.

<?hh // partial
//<?php

function main(): void {
  $ship = new Ship();
  $rf = new ReflectionFunction($ship->getClosure());
  $rf = $rf->getParameters()[0]->getDeclaringFunction();
  \var_dump($rf->isClosure());
  \var_dump($rf->getClosureScopeClass());
  \var_dump($rf->getDeclaringClass());
}

class Ship {
  public function getClosure() {
    return function(string $canon_to_fire) {
      return false;
    };
  }
}

main();

P.S. This doesn't need to be a PHP closure to have his behavior. I used one because of the <?php <?hh // partial dual operation of this script. This code will not work in the future if php closures are removed from the language. 4.30.0

[lexidor]

I am going over old issues on this repository, to see which ones apply to the current versions of hhvm.

We've come full circle.
I am closing my own issue about needing to close issues.

To all those issues that are on the list above, that I have not closed:
Sorry for marking your issue as one that needs to be closed.
With my far more conservative view on the issues list, I have closed some, but not all of them.
This issue was not made with the level of care that is required for working on open source.

I mainly focused on closing issues that relied on features that were removed from hhvm or relied on php mode.

I am updating the issue title of this issue to three underscores in order to not swing the discussion on whether to keep your issue open.