mini-ndn

Is this issue related to porting nfd integration tests? I don't see usage on the gerrit change.

Another thing is that result directory is made specifically for NLSR experiments - we should also remove that to make mini-ndn more general.

Ashlesh Gawande wrote:

Is this issue related to porting nfd integration tests? I don't see usage on the gerrit change.

Another thing is that result directory is made specifically for NLSR experiments - we should also remove that to make mini-ndn more general.

This is actually something which effects Mini-NDN-Wifi most directly, although now that you mention it would be excellent to use it with the NFD integration tests.

And agreed, I'll tweak that.

How is this used Mini-NDN-Wifi?
Can't we just store all the experiments in sub folders within ndn.experiments and recurse through it?
Where will the experiment manager be set - in bin/minindn? Would it be set from a command line argument?

Ashlesh Gawande wrote:

How is this used Mini-NDN-Wifi?
Can't we just store all the experiments in sub folders within ndn.experiments and recurse through it?
Where will the experiment manager be set - in bin/minindn? Would it be set from a command line argument?

In order to accomplish the task of separating out the Mini-NDN-Wifi and Mini-NDN libraries, which will decrease maintenance time significantly, we would necessarily need to have the Mini-NDN-Wifi files in a separate module which does not overlap with that of Mini-NDN. The downside is that experiment_manger presently hardcodes imports to be from Mini-NDN (https://github.com/named-data/mini-ndn/blob/master/ndn/experiment_manager.py#L42), which means that we are forced to create a duplicate (or wrapper class) in Mini-NDN-Wifi to change this one line. This adds maintenance burden for any changes made to experiment_manager or any other derivative code.

In regards to your second point, this would require redistributing Mini-NDN-Wifi experiments with vanilla Mini-NDN or redistributing modified Mini-NDN code with Mini-NDN-wifi (egg module files are compressed and cannot be easily altered).

To your third point, the former seems most likely. A command line option would be doable potentially but would need to be structured so it didn't cause regressions when interacting with other arguments.