Sort by end of word can't ignore diacritics

Question

A user just alerted me to this:

The Misaje languages have (as most languages in Cameroon) tone marks on vowels (grammatical marks to be more precise).
When you sort the wordlist by word, this works very nicely. It sorts the words regardless of tone marking, and then when there are multiple words with the same letters it will put these below each other:

When you sort the wordlist by End or word, the vowels with markings suddenly appear even under the letters that are not part of the alphabet (r, x), in the order of markings (first all low tones, then all high tones, …):

---

The language uses decomposed diacritics, so the ordering in the second screenshot makes perfect sense if it is taking diacritics into account. Unfortunately, “Ignore diacritics when filtering wordlist” has no effect on sorting, only filtering. It seems like it should affect the sorting.

This is how the alphabet is configured:

Answer 1

I’m not getting that result when I try it. My test project shows this when sorting from the end of the word: (not putting all the low tones first, then all the high tones).

Comments

@sewhite It looks like your example may be using composed characters (all of the letters are “normal” English letters that have composed equivalents). Can you check again with a project using decomposed normalization and some non-English letters with diacritics?