Mitigate slow kernel discovery (API adoption, placeholder kernels, measurements)

[DonJayamanne]

#5861

Problems

• Unable to run a cell after opening a notebook for first time after installing VS Code
  • Basically user needs until all Python interpreters have been discovered. This can be very slow depending on a number of factors.
  • The users perception is They cannot run a cell

Tasks/Suggestions:

• Review (& provide feedback) the new Python API
  • What will it take to implement this
  • Do we need changes in the API
• Progress indicator for kernels loading/being fetched (would be be of any value?)
• Add a placeholder lazy kernel for Python? (see corresponding issue with usecases for Lazy kernels)
• Add active interpreter as a controller (while discovering other interpreters)
• Add telemetry (if we don't already have this) to determine how quickly we return controllers
• Add telemetry (if we don't already have this) to determine how quickly we get interpreters from Python extension

[DonJayamanne]

[DonJayamanne]

@Tma2333 please could you file an issue with the problems you are facing. currently we do cache kernels for jupyter. please file an issue and we can continue the discussion there

[DonJayamanne]

Closing as this isn't requried anymore and all of the work has been done or no longer relevant.