I've been using QGIS Refactor fields to change type, re-order and rename a substantial ammount of columns (statistical data). Most of them hold purely numerical values and are as such not material to whatever decoding/encoding happens between various codepages, except for one. This column holds names (place names) which contain characters č, š and ž. Original data I procured is encoded in utf-8. For various practical reasons I exported them as cp1250 (default for all the computers involved) which produced exactly the same file with properly displayed symbols (čšž). When I applied Refactor fields everything was copied to a new file but č symbol was exchanged for c, while š and ž remained as they were.
I don't see any way to specifiy encoding in Refactor fields and since this is based on Python (which I know defaults to utf-8 in 3.x) means that I have encountered this kind of problem before. In Python that could be corrected by specifiying a codpage (in most cases).
What else could be the source of this issue and, in particular, can it be avoided using the command line version of the Refactor fields tool?