* All the file paths need a unique starting directory, so that events
from two different repositories don't have the same file name.
* If user names are not unique across two repositories, the structure
of the visualization will be heavily affected. The graph will contort
such that
the one author node connects to files from both repositories.
Perhaps this is what you are going for, perhaps not.
* If there are no common user names or file paths between two
repositories, there will be hardly any force interaction between the
two graph clusters. You will only have the slight repelling force
between file nodes. It's even possible one graph could be repelled off
the screen, as there is no center anchor, and the camera does not pan
to the center of mass of the system. (The only anchoring force is the
fact that files spawn at a random location, which means there is a
constant average force towards the middle of the screen as new files
pull on their submitters).
Note - I haven't followed trunk in months, I hope my underlying
assumptions still hold.
Also note - my company (www.krugle.org) has discussed having me setup
a weekly automatically generated video that tracks the last week of
OSS development across the top N most popular projects (for large
values of N). We crawl that kind of data anyway. So I hope
multi-repository visualization is both possible and visually
appealing.
-Alex