Implement readding of interposed questions

[luisa-beerboom]

Closes #2233

@jsangmeister I implemented it this far.
No idea if this is the wanted behaviour, I couldn't seem to find any meaning behind the weight distribution that gets created as a result of the various speaker actions. As a result I can only make guesses as to what logic I should be implementing here.

The action wiki entry is no real help either as the action entry for los.re_add_last doesn't seem to be up-to-date with the feature implementation (f.E. "The new waiting speaker gets the weight of min-1 of all waiting speakers or 1, if there are no waiting speakers." wherein the "or 1" part doesn't seem to currently be implemented).

If this doesn't work in the client, please tell me what exactly the result of the request should be

[jsangmeister]

Seems fine code-wise. Please enhance the docs to incldue the weight calculation for interposed question (that you do not decrement the weight). Not tested functionally.

Nevermind, I forgot that interposed questions are only possible for currently speaking speakers. See my next review.

[jsangmeister]

Please revert the changes to the weight calculation. Re-adding an interposed question should only be allowed if there is a currently speaking speaker (being_time != null && end_time == null). In this case, the previous weight calculation was already correct. With these changes, the rebuilding of the parent-child structure should work out-of-the-box.

[luisa-beerboom]

Please revert the changes to the weight calculation. Re-adding an interposed question should only be allowed if there is a currently speaking speaker (being_time != null && end_time == null). In this case, the previous weight calculation was already correct. With these changes, the rebuilding of the parent-child structure should work out-of-the-box.

Why would it be correct? That sort of weight calculation would place the interposed question above the parent motion, wouldn't it?

[jsangmeister]

Please revert the changes to the weight calculation. Re-adding an interposed question should only be allowed if there is a currently speaking speaker (being_time != null && end_time == null). In this case, the previous weight calculation was already correct. With these changes, the rebuilding of the parent-child structure should work out-of-the-box.

Why would it be correct? That sort of weight calculation would place the interposed question above the parent motion, wouldn't it?

No, they get the same weight, so the interposed question is sorted after its parent. You can try it with the current main: If you stop a speaker with an interposed question, then re-add the speaker and start it and then re-add the interposed question, everything works as intended. The only thing that must be prohibited is re-adding the interposed question if no one is speaking since then the UI breaks.

[luisa-beerboom]

The previous functionality literally sets everything to "weight": lowest_weight - 1. That's never going to be the same as the parent motion

[jsangmeister]

The previous functionality literally sets everything to "weight": lowest_weight - 1. That's never going to be the same as the parent motion

The weight is irrelevant for interposed questions, as they are always displayed directly after the current speaker. Only the relative weight between them is relevant, and for that purpose it is correct.

[luisa-beerboom]

@jsangmeister This is constantly failing it's checks. I don't know why. They don't seem to fail when I run the tests locally.
Do you have an idea what might be going on?

[jsangmeister]

@jsangmeister This is constantly failing it's checks. I don't know why. They don't seem to fail when I run the tests locally. Do you have an idea what might be going on?

It was successful now. I don't know why this happens sometimes, it seems to be a race condition in the migration route tests...