marbles | Read better test failures | Testing library

We figured it out. As you can see in our packages.json, we have a dependency on webpack. It seems angular-devkit/build-angular does as well. We believe this created the known issue of multiple webpacks colliding and causing issues. Removing our dependency on webpack fixed the issue.

Your main question is:

I would like more granular control of observable behavior. I don't really intend to complete the getStatus() observable.

If you do not want to complete it then it essentially is not a cold observable, its a hot observable, that would mean you can create a subject and pass

Some things to limit the recursion:

• If R>G+B+1, then there is no way to avoid having 2 adjacent reds. (Similar argument for G>R+B+1 & B>R+G+1.)
• If R=G+B+1, then you alternate reds with non-reds, and your problem is reduced to how many ways you can arrange G greens and B blacks w/o worrying about adjacency (and should thus have a closed-form solution). (Again, similar argument for G=R+B+1 and B=R+G+1.)

Apparently it was a bug

https://github.com/renovatebot/renovate/pull/9949

In the version 25.18.5 should be fixed

Since EMPTY completes without emitting next or error, you could use that as the test

Definition/lifetime of variables
Regarding this:

Most mind-boggling: the Check class does NOT have the global keyword in it, and it still works?

You only need the global-keyword for creating or changing global variables in a local context, see this answer.

In addition, the statement global app does not yet create the variable app - it is only created on first assignment. In your example, this means that app is only created here: app = MainWindow(window). If Question gets created before this statement (in your example, by Tk() or by MainWindow(window)), app does not exist yet.

You can try this out in an interpreter:

I've done the following now - I think it works, but I'll heave this open in case anyone can think of a more elegant way (or can think of an issue with my current implementation)

You would need to place the logic into a loop and implement a winning condition (given that you are somewhat cheating in deciding who starts, the computer will always win).

Also, your computer player should be checking for the modulo 4 of the number of remaining balls which you can verify using x%4 == 1 for example ( x == x+1 will never be True ). And, given that the computer will always play the modulo 4, you don't need several conditions, you can just use it directly.

You could take a simple mapping with a look for the wanted depth.