DOFs of higher order (>1) basis functions being ordered differently #84
Description
This is a repeat of the forum thread with the same name.
To summarize, the issue is that if a mesh is created, refined on a subdomain, saved, then loaded, the DoF ordering of order >= 2 H1 spaces on that mesh will no longer be consistent, which can be a significant issue if you have any code relying on said ordering (e.g. low-rank decompositions, as in my case).
Refining the entire mesh after the initial subdomain refine (before save/load) does return DoF order consistency, but this is often impractical.
I have attached a minimal non-working example. Please inform me whether I can provide any further information to address this issue - it is a breaking issue for my code.