It has been mentioned in the vignette of dendextend R package:
It is also worth noting that some order are impossible to achieve for a given tree’s topology. In such cases, the function will do its “best” to get as close as possible to the requested rotation.
I wonder if that is possible to retrieve the final order of the branches after rotating given I supplied some "impossible order"?
I am going to reorder a dendrogram with more than 20000 nodes (from WGCNA output), I found that genes were not clustered together while I believe it can be solved by rotating the branches.