gianm commented on issue #11929:
URL: https://github.com/apache/druid/issues/11929#issuecomment-974315460


   It sounds like PARTITION BY and CLUSTER BY are the winners! Works for me.
   
   I'm planning to put up a patch soon with the initial work to support 
planning INSERT (parts 1–3 and 5 from the "Proposed changes" list in the 
original proposal). I probably won't do any special parser stuff in that patch, 
since I think it'll get too complicated to try to do it all at once. So it will 
probably need to be a sequence of a few patches to get everything in place.
   
   > As for the type signature, I want to know do we have plan to support DDL 
such as CREATE TABLE? If DDL is supported, we can split the type signatures 
into a CREATE TABLE statement which would be more convenient for users to get 
started with Druid.
   
   I don't have a solid plan but I would like to support CREATE TABLE at some 
point. I imagine it would involve having a place to put datasource-level 
metadata like preferred partitioning, preferred schema, etc. I think it is a 
strength of Druid that datasources don't _have to_ have a strict schema, so I 
wouldn't want to enforce that, but I think it'd be good to let people create 
some schema if they want to.
   
   I also imagine that if you _did_ have a schema defined at the table level, 
then the INSERT command wouldn't necessarily have to always have a type 
signature, because in many cases it could be inferred from the table schema. So 
I think at that point we would want to make it optional.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to