I'd like to propose a generic interface for node data import (e.g. ARP caches) as an addition to regular nedi.pl discovery runs.
This would allow for an easy addition of information from devices which do not support the regular libsnmp and libcli collection
methods. This could be used for prototyping additions to the NeDi libraries as well as a permanent implementation.
A basic implementation would just import a file (say .csv) at the end of each nedi.pl run.
That file would have columns directly corresponding to fields in the nod* tables in NeDi.
Once in place, such an import mechanism could also handle node data which cannot easily be mapped to regular devices in NeDi,
think of SDN infrastructures, or leightweight virtual devices.
It might be necessary to add something like a pseudo device into NeDi in oder to handle this kind of data.