Coming in a little late to the conversation, thanks to phasefx for pointing me to the old messages. I'm also a little ignorant on the data and logic models of evergreen as I'm just starting to look deeper into it, so apologies for any dumb assumptions. I'm still working on a more thorough email about our experience but here's a little bit to keep the conversation alive.
In the simplest sense, roaming items are like an unconference, items vote with their feet and if they are there they belong. We roam high turnover items like dvds and cds though, and they seem to complicate things, especially with some of the return patterns that evolve. For example a good storytime can wipe out a chunk of a youth dvd collection only to have them all returned at another branch a week or so later when another popular storytime is held there. You would want returns from other locations to go to the wiped out location in the meantime to keep the shelves stocked. And then at the branch that the returns hit have more involved workflow to see what branches need items, have duplicates, have demand for the item, holds, etc. The home library may be full or have no real meaning in this sort of system. I guess this would be more of a free for all with the home location really being the starting point. Having the ability to have granular inventory caps and workflows could be really helpful in this case. I'm not supposing an easy interface to such a thing but having minimal places to modify the logic could be useful in implementing it on the code customization side. Overall I agree with the original post thought I think the example above shows that the balancing/inventory should have some importance. Again, I'll send a more detailed email on some use cases and issues we've ran into with our current setup in the future. Ryan Eby System Administrator Ann Arbor District Library
