DjangoFilterConnectionField ignores .using()

[dmalisani]

class ArticuloNode(DjangoObjectType):
descripcion = graphene.String(source='descripcion')
class Meta:
model = Articulo
filter_fields = {'nombre':['icontains', 'iexact'],
}
interfaces = (graphene.Node, )

class Query(graphene.ObjectType):
articulos = DjangoFilterConnectionField(ArticuloNode)

def resolve_articulos(self, args, context, info):
print(Articulo.objects.using("db").all())
x=Articulo.objects.using("db").all() <----- Output printer it's ok, but the def returns ignoring "db" and shows "default" db result.
return x

[japrogramer]

You might want to check that the DjangoObjectType doesn't alter the queryset.

[dmalisani]

How? I'm newbie in graphene
But I showed all code, there is no more my code between this code and requested for /graphql

[japrogramer]

I looked and couldn't find any change of the queryset that would change the queryset. Maybe the next person that sees this issue can be of more help.

[dmalisani]

if I only change
this line:
articulos = DjangoFilterConnectionField(ArticuloNode)
for this:
articulos = graphene.List(ArticuloNode)
it works perfectly (but I can't use filters)
I don't touch in anywhere out of schema.py and it works, then I concluded that It was a bug.
But, I still searching... Thanks

[dmalisani]

I think there is a bug in connection_resolver, I paste since line 89 of fields.py

    iterable = maybe_queryset(iterable)
    print("before ---->>  ", repr(iterable))

    if isinstance(iterable, QuerySet):
        if iterable is not default_manager:
            default_queryset = maybe_queryset(default_manager)
            iterable = cls.merge_querysets(default_queryset, iterable)
            print("after ---->>", repr(iterable))
        _len = iterable.count()
    else:
        _len = len(iterable)

Output shows:
before ---->> <QuerySet [<Articulo: Accesoriox Cable Canal()()()( - 14x7>, [...]
after ---->> <QuerySet [<Articulo: Accesoriox Cable Canal - 14x7>, [...]

It function evidently overrides queryset.

[japrogramer]

try doing
DjangoFilterConnectionField(ArticuloNode, on=True)
that should fix it.

[dmalisani]

Didn't work ---------------------------------------- File "[...]/graphene_django/filter/fields.py", line 96, in get_resolver self.get_manager(), File "[...]/graphene_django/fields.py", line 54, in get_manager return getattr(self.model, self.on) TypeError: getattr(): attribute name must be string [10/Aug/2017 18:17:02] "POST /graphql? HTTP/1.1" 500 177859 2017-08-09 21:15 GMT-03:00 japrogramer <notifications@github.com>:

…

try doing DjangoFilterConnectionField(ArticuloNode, on=True) — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#238 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFtT3nA5LCZDtBQ7LmfMhsKoNmpPU9Nyks5sWkujgaJpZM4OwvhZ> .

[dmalisani]

I commented that lines

        if iterable is not default_manager:
            default_queryset = maybe_queryset(default_manager)
            iterable = cls.merge_querysets(default_queryset, iterable)

And it works perfectly. I can't understand what is the use of that code.

[japrogramer]

Earlier i spoke incorrectly, see 'on' should actually be the name of the attribute for the manager that you want to use, so if you are using a custom manager to construct the query and don't tell the filter field it replaces it with the default manager. however i don't know if by passing on='objects' to the field will clear the using statement set later In the query.
sorry about that, i quickly looked over the code when i gave you that answer.

[BossGrand]

I think theres actually a bug with DjangoFilterConnectionField.merge_querysets

the line queryset = default_queryset & queryset

should be reverse to queryset = queryset & default_queryset

because default query set is overriding any attributes queryset would have. Such as order_by annotations ect. I'm about to push a pr to fix this issue