JSON-LD files for harvester and file that tells the crawler what to look for - how they should be named and kept up-to-date?

There should be one file that defines the files that have station information in them. lets say it is called sitemap.txt

This file should be always in the known location with the same name so the crawler can check it. At first time the file is populated with new filenames which all have station metadata in them. What happens when metadata changes at station?

- Should we remove the old filename from sitemap.txt and create new filename with the new data? Or should we keep the same filename and just change the data inside the file? Is there any restrictions for the naming of files? I have thought using hash function or some other random naming convention but if the names are kept the same, will use station names maybe.
- All this comes how the harvester is working. Does it always check the files for modifications or is it checking filenames for changes?

Tapani