Vscode-jupyter: python.showStartPage doesn't always work

[luabud]

Environment data

• VS Code version: VS Codespaces (aka.ms/codespaces)
• Extension version (available under the Extensions sidebar): 2020.7.96456
• OS and version: N/A

Expected behaviour

"python.showStartPage" : false will make the start page not show on updates/first time install

Actual behaviour

start page still shows _sometimes_ even though python.showStartPage is set to false.

Steps to reproduce:

[NOTE: Self-contained, minimal reproducing code samples are extremely helpful and will expedite addressing your issue]

1. Clone MicrosoftDocs/ml-basics in codespaces
2. Notice "python.showStartPage": false value in devcontainers.json
3. See the Python start page appear anyway

This isn't reproducible 100% of the times for me, though.

[LunarLanding]

I would like to add that a start page is redundant and invasive; there's already the extension homepage that the user forcefully goes through on installation. The fact turning it off doesn't work just makes matters worse.

[pewscorner]

I have the same problem, except that it always happens for me (either if a python file is open when the update happens, or the next time I open a python file after the update) - not just sometimes.

[greazer]

(internal: not strictly a data-science issue, but marking that way for tracking)

[DaelonSuzuka]

Been having this issue for weeks as well.

[IanMatthewHuff]

Sorry this was hanging around in the validate queue so long. So I do validate that the setting is getting set now, and with a forced downgrade and upgrade of the extension the start page didn't show for me. Validated.

However just to loop in @luabud quick on this since she would probably have the best take. If the setting is set to false to not show the start page you can use the command to pop up the page manually. This seems fine as the setting is just to prevent the autoshow and you can use the command to still bring it up.

But if you do this the checkbox for "Don't show this again" is unchecked, even though the setting is on already. I think that the checkbox might just be functioning as a button as opposed to actually reflecting the value of the setting. Do we care about that? I can file a bug if needed, but not really sure if it's worth it.