Re: [PHP] How do I handle covariant parameters and not fall foul of LSP.

On 13 February 2017 at 16:36, Stefan A. <[email protected]> wrote:

> I know it might look like deflecting the issue but maybe this could work
>
> abstract class BaseRepo
> {
>    protected function doProcessEntity(BaseEntity $entity){}
> }
>
> class PersonRepo extends BaseRepo
> {
>     public function processEntity(Person $entity)
>     {
>         $this->doProcessEntity($entity);
>     }
> }
>
> On Mon, Feb 13, 2017 at 5:13 PM, Richard Quadling <[email protected]>
> wrote:
>
>> Hi.
>>
>> I have an abstract base repo and an abstract base entity. Both provide
>> functionality
>> appropriate to all repos/entities.
>>
>> I extend these to allow a specific repo to handle a specific entity.
>> But I can't
>> declare the specific entity type hint in the specific repo as it is in
>> conflict.
>>
>> I'm hitting my head against LSP, but wanting
>> https://en.wikipedia.org/wiki/Covariance_and_contravariance_
>> (computer_science)#Contravariant_method_argument_type
>>
>> <?php
>>
>> abstract class BaseEntity{}
>>
>> abstract class BaseRepo
>> {
>>    public function processEntity(BaseEntity $entity){}
>> }
>>
>> class Person extends BaseEntity{}
>>
>> class PersonRepo extends BaseRepo
>> {
>>     public function processEntity(Person $entity)
>>     {
>>         parent::processEntity($entity);
>>     }
>> }
>>
>> $person = new Person;
>> $personRepo = new PersonRepo;
>> $personRepo->processEntity($person);
>>
>>
>> Warning: Declaration of PersonRepo::processEntity(Person $entity)
>> should be compatible
>> with BaseRepo::processEntity(BaseEntity $entity)
>>
>> I know some people 'get' this. But for me, it really seems to make next
>> to no
>> sense at all.
>>
>> I have some base behaviour for all entities. I have a base repo that can
>> work
>> with that base entity, regardless of how that entity is extended.
>>
>> If the base type were not abstract, then I can see that I wouldn't
>> substitute
>> anything logically, but I'm using abstract base types. And if set theory
>> is
>> to be believed, then if B extends A, B is still an A. And so, where I can
>> receive an A, I can receive a B.
>>
>> Obviously, I'm missing / lacking an understanding of LSP as it relates to
>> my
>> code.
>>
>> I'm happy to read more, but I'd appreciate some recommendations on what to
>> read.
>>
>> Of course, if anyone can 'correct' my code such that a concrete repo
>> works on
>> concrete entities, but also allows for the base behaviour, then I'd be
>> very
>> grateful.
>>
>> Regards,
>>
>> Richard.
>>
>
>
That's what I'm having to do. Just seems really odd.