. - CassandraVectorStoreAutoConfiguration uses CassandraAutoConfiguration - driver profiles for production stability+performance, - CassandraVectorStore.addStmts map must be concurrent-safe - code cleanups and naming fixes (field->column), - main doc tidy-up - astradb compatibility (protocol V4)
Comment From: michaelsembwever
A question: when calling vectorStore.add(docs) are all the docs expected to need embeddings created for them, or may this be optional ?
Currently all the stores are creating the embeddings in their add(docs) method.
But the spring-ai api does make it possible for the user to do the embeddings first and then call add(docs), shouldn't stores be honouring this…?
Comment From: michaelsembwever
when calling vectorStore.add(docs) are all the docs expected to need embeddings created for them, or may this be optional ?
got an answer from @tzolov , basically all in https://github.com/spring-projects/spring-ai/pull/413
adding a squash commit that provides the same functionality inside CassandraVectorStore (it's a one line change).
Comment From: tzolov
Hey @michaelsembwever , thanks for the great contribution. All looks good!
Comment From: tzolov
rebased, squashed and merged at 0eaf7d05c9bc509f8ae2d9e16594ec2db4fd2b4a