DjangoOptimizerExtension does not work on subscriptions

[rcybulski1122012]

Describe the Bug

A queryset returned from a strawberry_django.field resolver is automatically optimized when the type is resolved as part of a query or mutation, but not when the exact same type is yielded from a subscription — there it silently N+1s.

Minimal example

# models
class Color(models.Model):
    name = models.CharField(max_length=50)

class Fruit(models.Model):
    name = models.CharField(max_length=50)
    color = models.ForeignKey(Color, related_name="fruits", on_delete=models.CASCADE)

# types
@strawberry_django.type(Fruit)
class FruitType:
    id: auto
    name: auto
    color: "ColorType"

@strawberry_django.type(Color)
class ColorType:
    id: auto
    name: auto

# subscription payload — same shape that works fine as a mutation payload
@strawberry.type
class FruitsChanged:
    ids: strawberry.Private[list[int]]

    @strawberry_django.field(graphql_type=list[FruitType])
    def fruits(self) -> QuerySet[Fruit]:
        return Fruit.objects.filter(id__in=self.ids)

@strawberry.type
class Subscription:
    @strawberry.subscription
    async def fruits_changed(self, info) -> AsyncGenerat
        async for changed_ids in listen_for_changes(info):
            yield FruitsChanged(ids=changed_ids)

subscription {
  fruitsChanged {
    fruits { name color { name } }
  }
}

Expected: 1 query with select_related("color") (which is what happens if the same
payload type is returned from a query/mutation).

Actual: 1 query for fruits + 1 query per fruit for color - N+1

System Information

• Operating system: MacOS 26.5.1
• Python version: 3.12.8
• Strawberry version (if applicable): 0.319.0

[bellini666]

@rcybulski1122012 maybe we can use the new hook we have for subscriptions, to optimize the queryset, like we do for queries and mutations? That hook got added recently to strawberry

[rcybulski1122012]

I'm gonna take a look at it this weekend. Created an issue so it doesn't get lost 😄