The following should be guaranteed:
Example: NUMDAM [[13] ] uses a file naming scheme like
for journal volumes: journal-acronym_year_series_volume_issue for articles: volume-id_first-page_order
(different fields separated by '_'). This scheme allows to assign to any possible NUMDAM file (at logical units level: serial, volume, issue, article) a unique ID, making it robust and general.
Of course, conventions for file name length have to be obeyed.
Explanation: Since there will be numerous single DML projects, and since it seems too optimistic to expect that there will be a unique naming scheme for all projects/servers, a prefixing method identifying the project/server is recommended.