Symfony42: Remove invalid rectors

[bendavies]

Ref: symfony/symfony@f75fffa

These deprecation warnings are for when inherited class calls parent method.

[TomasVotruba]

I don't understand. The commit link is confusing.
Could you explain bit more?

[bendavies]

These deprecations warnings for for classes extending the symfony ones. for example.

This will trigger a deprecation warning.

<?php

class ClassThatInheritClient extends Client
{
    public function submit(DomCrawlerForm $form, array $values []))
    {
        return parent::submit($form, $values);
    }
}

It should be fixed to:

<?php

class ClassThatInheritClient extends Client
{
    public function submit(DomCrawlerForm $form, array $values [], array $serverParameters = [])))
    {
        return parent::submit($form, $values, $serverParameters);
    }
}

The current rectors are incorrrectly changing calls to the current symfony methods, i.e.
Changing this:

<?php


class MyService
{
    public function __invoke()
    {
        $client = new \Symfony\Component\HttpKernel\Client(...);
        $client->submit($form);
    }
}

to this

<?php


class MyService
{
    public function __invoke()
    {
        $client = new \Symfony\Component\HttpKernel\Client(...);
        $client->submit($form, []);
    }
}

[TomasVotruba]

I see. In that way it should not be removed but fixed.

There should be new configuration option, that adds it only to parent call + class method.

 services:
     Rector\Rector\Argument\ArgumentAdderRector:
         Symfony\Component\BrowserKit\Client:
             submit:
                 2:
                     name: 'serverParameters'
                     default_value: []
+                    type: ['class_method', 'parent_call']

[bendavies]

👍 so it's not possible right now?

[TomasVotruba]

Not yet

[bendavies]

Right. So shall we merge this for now? As this config is causing incorrect changes

[TomasVotruba]

No, it needs to be fixed so it works as it should.
Merging would cause other troubles with missing migration

[bendavies]

up to you, but at the moment keeping this causes an incorrect migration, which is better than no migration.

[TomasVotruba]

I know. But finished work is better than double work.