[ZEPPELIN-3368] Refactoring code completion (plugins)

• Type: New Feature
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: interpreter-auto-completion
• Assignee: Maksim Reshetov
• Reporter: Maksim Reshetov
• Created: 27/Mar/18 09:44
• Updated: 28/Mar/18 07:45

Description

Create new entrypoint (plugin) for generate metadata. in the context of the task, create an interface for metadata generators and create a UI to manage them. Metadata must be stored in memory and generated immediately after the start of the server (+updated periodically),
what problems can be solved:

1. code completion is too slow (because generates in runtime), code completions more than 1 second it is not good
2. code completion does not work until the first start of the interpreter (initialization), if interpreter is 'per user' then it is awfull
3. if the interpreter (e.g. JDBC) is "per user" than we have more duplicates of metadata. And JDBC driver loads meta by method getMetaData, if data  is large then it is too slow

 

extra benefits:

  one entrypoint with all meta. posibility to create visual plugins for easy navigation and viewing of available data