From what I have seen, _index and _parent_index change as new data is submitted. This means that for data within repeat groups, there is no unique reference for that survey that is constant, so if I try to reference a row today you cannot find that row tomorrow. The only way to fix this is to manually generate your own uuid’s within every repeat group.
I don’t think this should be the expected behavior, as you naturally want to use an index to index your data, but it seems to have no use at all in its current state. Maybe I’m missing something, so what’s the reasoning behind the current behavior? Is it possible for KoBo to instead maintain a permanent index that doesn’t change once data is uploaded to the servers, something like indices based on submission time? If there is logic behind how it currently works, could we add an option to a project to have fixed state indices?
Thanks for any responses.