Improvements to getting started experience for DS

[DonJayamanne]

Should we consider having some kind of a workflow that'll help users setup VSC for DS work.
E.g for if Jupyter doesn't start and this is the first time a user has attempted to run jupyter then:

• We recommend installing conda (if not installed)
• We check if conda has been setup correctly
• We suggest picking a jupyter interpreter from a conda environment...
• I.e. we walk the user through the process..

Today, we ask user to install jupyter if not found, then later we ask user to install ipykernel..
Basically we have prompts with solutions, but only when users run into an issue.
I think we can improve this experience, rather than letting users stumble around and fix for themselves..

Kind of a Wizard approach, using standard prompts with steps like in Azure where they have Step 1, Step 2,, etc)

We could bring this prompt (that kicks off this workflow), when jupyter doesn't start or kernel fails, or the like.

[rholland]

I think this is a good idea. I have installed vscode-python on a dozen different computers over the last year, mostly Windows 10. They are all a little different - some Anaconda - some Windows store Python - some python.org Python. I don't do anything special but I may choose different options provided by the install routines. In almost every case "jupyter notebook" works from the command line but the interactive notebook/window in vs code does not. It typically takes 1/2 to 2 hours of googling and messing around to get it to work, and the reasons are always different.

[DonJayamanne]

@rholland thanks for your feedback, would it be possible to list some of the issues you encountered and how you fixed them. E.g. you learnt x was not installed (an error message was displayed that clearly indicated this) and you installed it manually (or the extension gave you the ability to install x).
Thanks

[rholland]

Sure. I will add things over the next week as I get a bit of time. Thanks very much for all of your good work. VSCode-Python is a pleasure to use.

[rholland]

The recent Jupyter initialization experience is getting better. From my point of view this issue could be closed. Thanks for the good work.