Daniel Gellert

@thePermission did you experience this in a later version? We used this in v 1.3.37 with another cube and it seems to work there... @igorlukanin is this silently fixed?

Can you elaborate what your usecase is? It seems like currently the serializers are guessed from typ, which can be either bytes, str, faust.Record, or faust.ModelT custom class. What is...

Wow thanks for the detailed description. According to the [code of faust the serializer type is guessed](https://github.com/faust-streaming/faust/blob/master/faust/tables/base.py#L177), and you can specify it on the model you define. The following very...

True, in case of pydantic I am not sure if it is allowed to define an _options attribute that also has an serializer attribute. That way you could peobably trick...

Actually yes there is, it basically works the same as pydantic v2 as well model_dump in pydantic dumps a python dict preserving the types of nested attributes, model_dumps_json on the...

I am not sure if I know what you mean. I have the following program: ``` import logging import faust logger = logging.getLogger(__name__) class Bar(faust.Record): baz: str class Foo(faust.Record): bar:...

Same for me with different files. Using python 3.9.7 and faust-streaming==0.10.16

python 3.10.9 same result on mac OS

@cmartmos I have not used the group_by functionality my own. While having a quick glance at the function there is a topic parameter. Can you create a topic and pass...

The thing is that test_context() wants to add a sink and this would trigger the check that we need to yield in the agent for a sink to be valid....