Fix error of multiple inputs with the same type. When using same seri…

[Hispar]

…alizer.

Fix #478

[coveralls]

Coverage increased (+0.04%) to 92.755% when pulling 97461d1 on Hispar:bugfix/multiple_model_serializers into b0cba39 on graphql-python:master.

[danpalmer]

I'm afraid I don't have any experience with Django Rest Framework, so I'm going to opt-out of reviewing this one.

[jkimbo]

I don't have any experience with Django Rest Framework either unfortunately.

[zbyte64]

Will this change require existing code bases to update their client? If I understand correctly the input type label will change.

[Hispar]

Will this change require existing code bases to update their client? If I understand correctly the input type label will change.

I'm not sure, the types are generated every time the schema is created. I have been using this fix several months without problems.

[dani0805]

We ran into this problem. The pull request solves the problem, but is it necessary to create 2 Input types? would't be more appropriate to reuse the first one that gets created? After all if its the same Serializer, should be also the same serializer input type right?

[Hispar]

We ran into this problem. The pull request solves the problem, but is it necessary to create 2 Input types? would't be more appropriate to reuse the first one that gets created? After all if its the same Serializer, should be also the same serializer input type right?

The correct way should be to reuse the first type as you are saying. But when I encountered this problem I didn't find the way to do it, so I implemented this workaround meanwhile. I haven't had time to look for a better solution.

[dani0805]

Yes, I hear you ... we commented the assertion :-)

something like this? (Did not test it as I don't have the project setup)

def convert_serializer_to_input_type(serializer_class):
    cachedType = convert_serializer_to_input_type.cache.get(serializer_class.__name__, None)
    if cachedType:
        return cachedType
    serializer = serializer_class()

    items = {
        name: convert_serializer_field(field)
        for name, field in serializer.fields.items()
    }
    retType = type(
        "{}Input".format(serializer.__class__.__name__),
        (graphene.InputObjectType,),
        items,
    )
    convert_serializer_to_input_type.cache[serializer_class.__name__] = retType
    return retType
convert_serializer_to_input_type.cache = {}

[Hispar]

Yes, I hear you ... we commented the assertion :-)

something like this? (Did not test it as I don't have the project setup)

def convert_serializer_to_input_type(serializer_class):
    cachedType = convert_serializer_to_input_type.cache.get(serializer_class.__name__, None)
    if cachedType:
        return cachedType
    serializer = serializer_class()

    items = {
        name: convert_serializer_field(field)
        for name, field in serializer.fields.items()
    }
    retType = type(
        "{}Input".format(serializer.__class__.__name__),
        (graphene.InputObjectType,),
        items,
    )
    convert_serializer_to_input_type.cache[serializer_class.__name__] = retType
    return retType
convert_serializer_to_input_type.cache = {}

It could work, at this moment I do not have the project configured to test it, but as soon as I can I will see if it works and I will update the PR accordingly.

Thanks!

[Hispar]

@dani0805 I have updated the PR with your solution, I haven't tested it in our projects yet, but the test of the schema works.

[kimbriancanavan]

This would solve headaches for us. Let's get this merged 👍

[Shehab-Muhammad]

why no one reviews and merges this, please merge this ASAP

[phalt]

Hey @Shehab-Muhammad please appreciate that this project is worked on my volunteers in their spare time!

[Shehab-Muhammad]

@phalt I appreciate your work, of course. thank you