Weaviate <-> Langchain : Would love some clarity

[EnviralDesign]

here's some sample code I pulled from the python wiki for langchain (0.0.179)
https://python.langchain.com/en/latest/modules/indexes/vectorstores/examples/weaviate.html

WEAVIATE_URL = "http://xxx.xxx.x.x:8080"
client = weaviate.Client(
    url=WEAVIATE_URL,
)

client.schema.delete_all()

loader = TextLoader("C:/tmp/state_of_the_union.txt", encoding="utf-8")
documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
docs = text_splitter.split_documents(documents)
embeddings = OpenAIEmbeddings()
db = Weaviate.from_documents(docs, embeddings, weaviate_url=WEAVIATE_URL, by_text=False)
query = "What did the president say about Ketanji Brown Jackson"
docs = db.similarity_search(query)

If you follow the instructions for installing weaviate here:
https://weaviate.io/developers/weaviate/installation/docker-compose

You can use their configurator to generate a docker compose file, which will successfully spin up weaviate. There are several options for ML modules you can have paired with weaviate, one of which is the openAI embeddings API.

I started off by dismissing this and going for a "bring your own vector" strategy as weaviate suggests is possible in their docs.
I assumed this was the correct way to go about it for integration with langchain as all of the guides show instantiating a langchain embeddings wrapper, and passing that into a vector store langchain wrapper.

This works for other vector stores, but does not for weaviate, there have been a few other issues surrounding this:
#2820
#4742

I am able to make weaviate function with langchain by setting up the openAI embeddings module in the weaviate docker compose file.. but that begs a couple of questions:

1. When weaviate is configured to use it's own embeddings module, and I also pass in an embeddings module from langchain.. which one is used?
2. If the openAI embedding specified in weaviate takes precedence internally over langchain's, and the langchain embedding is a meaningless argument, is there a work around to getting weaviate to use langchain embeddings?
3. weaviate has some really intriguing things going for it like hybrid search and others, is it possible to use hybrid search if the embeddings model from langchain is in fact being used?

Thank you!