Automatic population of references #10
Description
I wrote yesterday a script that can be used to populate automatically reference tiers. It works now perfectly when the file has no references and the idea is to add them. Basically it does the same as labeling the tier in Elan, so it replaces all existing content.
However, I wonder what to do with the cases where there are just missing or problematic references, so that some of them are empty or duplicates. Should only the missing one be replaced, and if, should the number be first non-used id or should it be what comes consecutively before it, and update all that are after it by one.
So do we want this:
To turn into this (this is what the script essentially does now):
Or this:
The first option has plus that it makes minimal change to ELAN which is easy to verify with git diff and similar tools, but it breaks the consecutiveness of the annotations. I don't know if this matters anything. One approach could also be just to rewrite all reference annotations every time there is need for it, but this also means the file always has as many changes as there are annotations, even though in reality the change would be only about few cases. In practice this would also change only annotations after the changed one, as the number of earlier ones would not change.