Poscar file not ordering correctly with respect to defects; U values then out of sync

[cmgtam]

How does Mast order elements in the Poscar when doing defect substitutions? I've just noticed a big mistake with a lot of my runs because although it directly gives the right pseudopotential, my U values and moments are wrongly assigned because the element ordering doesn't seem to follow a pattern (I thought it was alphabetical when doing defects).

For example, LaCrO3 base material, and alloying Cr site with Mn gives the following element order in Poscar: La Mn Cr O Doing LaTiO3 as a base material, alloying Ti site with Mn gives the following element order in Poscar: La Ti Mn O

In each case I'm substituting the same lattice sites and the same concentration. This makes it hard to tell which U values, etc. to use if you don't know beforehand how the ordering will go. Or, is there a way to directly assign the element tag the appropriate U value as you do with pseudopotentials, e.g. Mn = 3.9, Cr = 3.7, etc rather than using the typical 0 3.9 3.7 0 form that requires knowing the element order?

[cmgtam]

User input file had all of these elements as substitutional defects, after the initial structure creation. Because these are all created as defects instead of the initial structure, we actually do sort them in MAST with pymatgen's get_sorted_structure, which apparently sorts atoms by electronegativity. See ingredients/pmg_extend/structure_extensions.py, method "induce_defect" which goes to get_sorted_structure()

So that is why Ti would show up before Mn, and Cr would show up after Mn. This issue is yet another compelling argument for resolving Issue #385