I have a really weird error/data mistake on one of my cluster outputs. Unfortunately, I cannot include the file, but I appreciate it if anyone could point me in the right direction. I’m a bit baffled by this one.
So it’s a cluster within a cluster, and one instance of 7. I already tried disentangling with no results.
Ok, I think I know the reason. I kinda figured it out while writing the topic, but I thought I’d still post it if someone has similar issues in the future.
The reason seems to be the different handling of empty branches by the active document and the computation inside a cluster. Active document (including an opened cluster) somehow handles the irregular tree structure differently from the clusterised and hidden computation.
The error actually happens inside and after a python script component, so this could be an issue with the python computation behaving differently inside and outside of a cluster. Hard to say, but this one I solved by flattening before cluster, and the un-flattening afterward.
When I expect empty branches to enter a cluster, I like to split the tree and only move filled branches into the cluster. This way debugging/maintanance becomes much less of a hassle. I am interested in whether you found out that the python script actually breaks due to being in a cluster, or whether mistakes are explained by the mismatching of trees due to empty branches also entering the cluster? If the python script actually behaves unexpectadly, this would be very good to know for future reference!
@Gijs_Jonkheer
I have not found out anything related to Python component not working inside a cluster. I try to avoid the use of clusters altogether, but as far as I know, my encountered unexpected behaviour instances can be explained through empty branches.
