[WIP] Implement jupyter server listextensions

[echarles]

This PR implement the extensions listing as a subapp of jupyter server.

[vidartf]

So, where we previously did jupyter serverextension list, we now do jupyter listextensions? What about the install commands?

[echarles]

Hi @vidartf , This implements jupyter server listextensions instead of jupyter serverextension list.

The last one is still supported for backwards compatibility reasons. This PR is just a first shot further to a discussion I had with @Zsailer. We said we would further see how we should progress with that.

I have added a WIP to the title as we first need to define how we progress, especially on how we can make a distinction between the classical extensions and the new application extensions. I expect confusion for users and even developers...

[dhirschfeld]

listextensions seems a bit awkward and unnatural - it's not an actual word itself but a concatenation of two separate words.

Granted, the same could be said about serverextension but somehow it seems even worse. Maybe it's the combination of the verb and noun together instead of jupyter thing action it's jupyter thing action-on-sub-thing i.e. it's trying to cram a 3-level hierarchy into only two levels by conflating the last level with the action you want to perform on it.

i.e. jupyter has a server which in turn has extensions for which you want to perform actions such as list. Whilst a mouthful, I think the below might be a better api:

jupyter server extension list

If there are more actions that you want to support (install/remove/enable/disable?) then the above seems more composable than jupyter server installextension etc...

[rolweber]

If it is a technical problem to pass three command words instead of two, either of these would be more readable:

• jupyter server-extension list
• jupyter server extension-list

[Zsailer]

I agree with the comments above. I've opened a #157 as an alternative approach, implementing a jupyter server extension ... subcommand syntax.

[echarles]

Closing this in favor of #157