Conversation
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
samsja
left a comment
There was a problem hiding this comment.
Lets discuss I don't think we should implement this that way.
If I am understanding correctly here the json representation of a DocList and a DocVec is the same right ?
IMO DocVec should have its own column based json representation.
For csv etcc I would say we ask first user to instantiate a DocList then cast to docVec
Fair
Why? The whole point of this issue is to natively enable these, no? Why do you think DocVec should not have these methods? |
My point is that saving line by line in csv for instance is a bit weird for a column base format. But maybe yes we can have them, I just don't want it to be confusing |
yeah this is fair, I will address that |
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
Signed-off-by: Johannes Messner <messnerjo@gmail.com>
|
📝 Docs are deployed on https://ft-feat-docvec-io--jina-docs.netlify.app 🎉 |
This PR implements the IOMixinArray for DocVec, enabling the following (de)serializations:
Depends on #1561 (protobuf serialization bug)
JSON data now contains four columns, like this:
CSV is not supported, since it is a row-based format.
All other formats are used the same way as in DocList, but have a different serialized representation.
Closes #1330