Thanks Glynn and Eric for clarifying the links. That makes sense now.
It also gave me some ideas for how to handle modifications to support job tracking in grass. All our working directories (should) have a "jobinfo" file which contains client and project information. In-house software uses this to tag accounting files so that I can generate month end reports on projects or billing for clients. This is particularly important for cluster usage - but I also use wrappers on the sequential code so that distributed runs are tracked in the same system.
I'm curious if anyone else has set up distributed grass modules. I've been contemplating creating a user "grass" across the cluster which configures their environment at logon as if grass is running. Then if I rsh/ssh something on another machine... Alternatively there would need to be a version of front.end that handles remote invocation.
Cheers,
John Harrop