Kaapi platform API Versioning

Right now our API is effectively “latest-only.” That makes iterative changes risky and forces clients to upgrade immediately. We have situations where even though we need to evolve a feature's capabilities then we have to think of it in a way it does not affect our current users, this causes us to be limited with the changes we need to do in any module. To solve for this, we need a versioning strategy to:

• Establish a clear, documented API versioning scheme

• Stand up v1 as the first frozen version

• Enable development of v2 (and beyond) without breaking v1

• for example collection module creates an openai assistant from documents given to the module , but since assistants is going to be deprecated, we will make changes in the module by removing the assistant part and just keep the creation of vector store. although dalgo needs collection to create assitant that is why we will version the collection module where v1 will be the current version, and v2 will be when we will just be creating a vector store from the documents

Commencez par cartographier l’API actuelle du module collection et son intégration avec Dalgo, en vous concentrant sur le comportement de OpenAI assistant et vector-store décrit dans l’issue. Définissez ce qui doit rester dans v1, ce qui passe à v2 et comment le schéma de versionnage et la documentation montreront que v1 reste compatible tandis que v2 supprime la création d’assistants.