Component re-registration warning when two paginated list endpoints use the same serializer

[Lucidiot]

Describe the bug

When two or more paginated list endpoints use the same response serializers, a warning appears when generating the schema:

Warning #0: trying to re-register a schemas component with name PaginatedFooList. this might lead to a incorrect schema. Look out for reused names

To Reproduce

Create two list endpoints with the same serializer, and set a default pagination class in the settings (or add one in pagination_class on each endpoint).

class FirstList(ListAPIView):
    serializer_class = FooSerializer
    ...

class SecondList(ListAPIView):
    serializer_class = FooSerializer
    ...

Expected behavior

The same component gets reused without a warning.

Maybe this check could be modified to not warn when the two components are the same, not just by key? Or should this part check whether the exact same component already exists before registering?

[tfranzel]

hey @Lucidiot, yes that looks like a bug. the warning is unwarranted in most circumstances. i'll refine that spot to catch problems and not emit false positives.

[tfranzel]

that should do it. please confirm and close the issue

[ShaheedHaque]

I'm not sure if you would consider this the same issue, but (using the 0.12.0 release), I see this:

Warning #0: Encountered 2 components with identical names "PaginatedEmployeeList" and different classes <class 'api.employee.EmployeeSerializer'> and <class 'rest_framework.serializers.ListSerializer'>. This will very likely result in an incorrect schema. Try renaming one.

even though the ListSerializer (implicitly, via many=True) specifies the EmployeeSerializer as its child:

@extend_schema_view(
    ...
    employees=extend_schema(..., responses={HTTP_200_OK: EmployeeSerializer(many=True)}),
    ...
)

Or perhaps I am not using extend_schema as expected?

[tfranzel]

hey @ShaheedHaque, i think you are doing it right and i missed a case. the test case did only cover multiple viewsets and not actions. i can see how this happened. i'll investigate. the schema however should still be correct i believe.

[tfranzel]

@ShaheedHaque i extended the test. should now behave as expected. schema was unaffected, only a false positive warning.

[ShaheedHaque]

Excellent. And, as a newbie to (both drf and) drf-spectacular, I'd like to say thank you for the excellent software.

…

On Sun, 20 Dec 2020, 13:05 T. Franzel, ***@***.***> wrote: @ShaheedHaque <https://github.com/ShaheedHaque> i extended the test. should now behave as expected. schema was unaffected, only a false positive warning. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#233 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABFHWKMA4C2PXA62FZ32MVLSVXZBNANCNFSM4U5ZRPYQ> .

[Lucidiot]

Just updated and I no longer see any warnings. Thank you very much!