I'll preface this by saying I'm not sure if I found a bug or because I'm somehow misusing IrisVectorStore.

Basically I have code from the regular llama-index module working in my Python project which has SimpleDirectoryReader objects similar in nature to the demo I mentioned (https://github.com/intersystems-community/iris-vector-search/blob/main/d...). And I have other code working that can add new users to a SQL table in Iris.

I tried to use IRISVectorStore in a manner similar to the below excerpt from the demo code but I just changed the table name to the name of my user table. And I also just changed the documents object in that code to my own SimpleDirectoryReader object.

However no matter how many times I try to run with those changes I get a flurry of exceptions where the trace makes little sense to me. I can confirm that my code in place to connect to my user table locally does work.

I won't attach the trace yet unless someone asks but my question is basically does anybody know for sure that IRISVectorStore can successfully extract information from a user table? Or might I have hit some weird edge case when trying to use this?

# StorageContext captures how vectors will be stored
vector_store = IRISVectorStore.from_params(
    connection_string = url,
    table_name = "paul_graham_essay",
    embed_dim = 1536,  # openai embedding dimension
    engine_args = { "connect_args": {"sslcontext": sslcontext} }