Port 7.x changes to master

[jasongrout]

This merges 7.x into master to pick up changes since we branched 7.x.

Fixes #3085

The basic strategy here was to do a git merge 7.x on master, then fix the conflicts with these ideas:

• bump any internal javascript alpha version numbers to the next highest major version, to account for the major version release done on the 7.x branch
• ignore whitespace changes, then run lint at the very end to do formatting cleanups
• copy over the existing jupyterlab manager ts files from master over the 7.x jupyterlab_widgets package, after making sure that relevant changes from 7.x on those files were already incorporated into master (see the todo item below for some things I noticed doing this)

Checklist

• Make sure all of the changes to the jlab manager manager.ts file from the 7.x branch make it into the master branch: fdd7109...7.x#diff-fcef124ae167cb55c01a19fcf5edc6c00f4c49cbb4b56386f8acb3b434aaa386
  • Most of these changes were actually backported from Package upgrades #2654, which was merged into master
  • There may be more to do with how this._kernel works, like making sure it is actually set when the session context changes the kernel, but perhaps on another PR. This was written in 9d999d7, fyi. I've made Check logic around changing kernels in JupyterLab manager #3088 to examine a few items to check into later.
• Update https://github.com/jupyter-widgets/ipywidgets/pull/3086/files#diff-c0c3f9fac78e3426a2d3076dc6f327c613b69178c236bf8b7c478f59aac774d9 (for example, replace jqueryui license with noui slider license)

[jasongrout]

I think this is ready for review. I used dev-install.sh to install into a fresh environment and ran most of the doc example notebooks in JupyterLab, and everything looked fine to me.

Once it is reviewed and merged, I can release another 8.0 alpha.

CC @SylvainCorlay

[ianhi]

Just tried this out. Everything except for the Lorenz example worked great. I'm not sure if that's an issue with the forward port, or with master though, investigating.

[ianhi]

The failure above is unrelated - see #3102

I didn't read over all the code here but I tried it out and didn't run into any issues. I also checked and ipympl will run on jlab3 without issue when using this PR so custom widgets may not have so rough of a time 🎉

[jasongrout]

Thanks Ian!

I'll leave this open until tomorrow in case @SylvainCorlay or @vidartf wants to look at the code too.

[davidbrochart]

ipyleaflet works fine in jlab3 with this PR too.

[vidartf]

I tried creating a few branches to see if there were any history problems, but could not find any. I guess the only point of note is that the history of a few files will be a bit "back and forth" due to the cleanup commits, but I guess that should be ok.

[marimeireles]

Hi @jasongrout, I'm faar from being an expert in ipywidgets, but @SylvainCorlay asked me to check this PR. So, I'll give it a try and run some tests to see if I find something obvs wrong. I'll start now, I can ping you once I'm done.
Maybe @davidbrochart is also taking a look on this?

[marimeireles]

Oops, sorry, had an old page loaded here.

[vidartf]

Hmm. Inspecting the git history of a branch where this is merged into master using a merge-commit, I see one concerning item: This PR will cause tagged commits to be interleaved in the master branch. I'm also seeing a pretty complicated history-tree when using gitk. As such, I'm wondering if this should be merge back into master as a squash commit ? That will make it a little more tricky to navigate the history using git blame though ...

[vidartf]

This PR will cause tagged commits to be interleaved in the master branch.

Actually, no, it will only be interleaved in a chronologically sorted history view.

I'm also seeing a pretty complicated history-tree when using gitk.

This might be a good thing, not sure. My git skills aren't up to par.

[jasongrout]

@vidartf - I think on the whole it is good to have the merged 7.x commits as ancestors of master. That will help in, for example, the milestone check script over at #3091 to understand which commits are in 8.x but not in 7.x, etc. It also reflects in the history the actual process of merging in the changes

[jasongrout]

I'll start now, I can ping you once I'm done.

@marimeireles - great! I'm ready to merge this (or have someone else merge it), so I can't wait to hear your thoughts before doing so. I plan to release an 8.0 alpha after merging this so we can have a clean base to test and merge new things into.

[marimeireles]

Hey @jasongrout I tested with ipycytoscape and a few other smaller things. Everything worked for me :)

[jasongrout]

@marimeireles - thanks!

@willingc - this is merged now. I'm planning to make another 8.0 alpha release later this morning or early afternoon. Please let me know if you want me to hold off for one of the docs PRs you've been blocked on.