wp_die() and PHPStan

[jcvignoli]

I haven't found any phpstan specific instruction in the stubs, so I would like to ask you wether it's ok to add the following comment for wp_die():

 * @phpstan-return ($args is array{exit: false}&array ? void : never)

Without it, PHPStan doesn't understand that wp_die() is an exiting function.

[marekdedic]

If you keep the non-phpstan-specific return as well (so that other tools keep working), then adding this is absolutely ok.

2 questions:

1. Why is there the &array? Seems confusing to me
2. While we are poking wp_die, maybe the default return should be changed to never? I suspect for 99% of uses it's what they want...

Thanks!

[jcvignoli]

Good!

1. I Have no idea actually. Just copied-pasted from another stub, and frankly I'm not an expert about phpdoc definitions. I just tried
   * @phpstan-return ($args is array{exit: false} ? void : never)
   which worked out too.
2. I fully agree! I didn't want to mess out things, didn't know if was ok to use never (which is never used in your stubs).

I'm gonna make a PR with a @return never. Thanks!

[jcvignoli]

Well, actually I commited a @return never but PHPStan check fails:

 1372   Function wp_die() should always throw an exception or terminate  
         script execution but doesn't do that.                            
         return.never      

[marekdedic]

2. didn't know if was ok to use never (which is never used in your stubs).

I think if it passes CI...

Well, actually I commited a @return never but PHPStan check fails:

I think putting exit(); in the implementation will fix this. Or disable this rule in PHPStan config, it doesn't really add much for stubs anyway...

[jcvignoli]

I think putting exit(); in the implementation will fix this.
Putting an extra exit() in the code following wp_die()? Well, it wouldn't make sense, right?

PHPStan rule can indeed be disabled, but it would disable also proper issues at the same time.

I therefore used the @phpstan-return option, which passes CI.

[marekdedic]

PHPStan rule can indeed be disabled, but it would disable also proper issues at the same time.

Hmm, that makes sense... And putting exit() inside the function doesn't work either?

[jcvignoli]

Adding an exit() after a wp_die() would throw errors in other static tools (something like the code could not be executed), I guess.

Closing due to the discussion in PR.