Gereon Kaiping

This in turn seems to happen because the SA-Operators (I found it in [SAWilsonBalding](https://github.com/CompEvol/sampled-ancestors/blob/ff14640669dfb35acfdf52b814f489bb2d7aa16b/src/beast/evolution/operators/SAWilsonBalding.java#L156-L164), at least) first assign a new node as the root and only then add the remaining...

They are both required, but they aren't enough yet: I also ran into ``` Fatal exception: null java.lang.NegativeArraySizeException at beast.evolution.tree.Tree.store(Unknown Source) at beast.core.StateNode.startEditing(Unknown Source) at beast.evolution.tree.Tree.startEditing(Unknown Source) at beast.core.StateNode.getCurrentEditable(Unknown Source)...

I have put a [pull request through to the SampledAncestor repository](https://github.com/CompEvol/sampled-ancestors/pull/19), as well.

The issue with the SAWilsonBalding operator is that it constructs a partial tree, set that as the root, and only then glues in the other part of the tree, so...

> So, if the SAWilsonBalding operator would call `setRoot` with > the correct root at the end of the proposal, that would do the > trick, right? It should. I...

>> I see how using the `-1` flag works, but it does not feel particularly robust. > > Oh, it really is not. I already ran into it being a...

Yes, there's probably a lot of confusion in my head about these things. Thank you very much for engaging nonetheless!

I ran into practically the same thing today, with the error message given as ``` { "errors": "{'metadata': {'related_identifiers': {1: {'relation': [u'Not a valid choice.']}}}}" } ``` The root cause...

"isVersionOf" is indeed not listed as a valid relation under `related_identifiers` in https://developers.zenodo.org/#representation

If you just merged the `raise "string"` differences, 2to3 should run through without problems, and you don't need to keep two branches up to date, I guess. Sorry I did...