pocolibs

Currently, remotePosterLib can be used in two ways:

• POSTER_PATH = host1:host2:...
• POSTER_HOST = single-host

The choice of either of the two setups may depend on to which role you want to assign the penalty of the network latency:

• With POSTER_PATH, you set the network penalty on the reader (remotePosterRead() will lag, localPosterWrite() won't).
• With POSTER_HOST, you set the network penalty on the writer (remotePosterWrite() will lag, localPosterRead() won't).

It might be useful to be able to use POSTER_HOST with a host list, so that you can at the same time set the network penalty on the writer, and handle mutliple reader hosts by publishing a poster on all of them in a row.

Of course, the question of "what happens if a remotePosterWrite() fails in the middle of the host list?" is relevant. The simplest is to just skip the posterWrite() for that host, continue with the remaining hosts and finally return an error to the caller. One of the remote reader will not see the updated version, and have no hint about this. But this is no different from the current situation with a single host in POSTER_HOST: if remotePosterWrite() fails, the remote host is not aware of it.

In practice (real word example), this would be useful for broadcasting a "motion capture" poster (optitrack-genom3) from a ground machine to multiple wifi machines, withouth impacting the readers (pom-genom3) on those remote devices. Currently, the only way to implement this is to set POSTER_PATH=ground-machine on each wifi device, which will delay the reader task of pom-genom3 and prevent it from timely reading other (local) data sources.

If handling POSTER_HOST=host1:host2:... sounds OK, I could work on it.