old thread, but regardless....
my concerns:
1. hundreds of fields within window manager will get messy. remember that it's possible to cause irreversible damage to a window in rare cases. i have seen weird things happen to windows when i'm being impatient. (starting at bootcamp) if you do this, I would keep some copies of the windows before you mess with them.
2. another thing to consider is that one giant process means all your eggs (users) are in one basket. This sounds like a dream for alterations for repetition avoidance, but it can be an even worse nightmare. should you need to take it offline (even for something tiny) you have to take down every process. it also prevents you from selectively rolling out a new version of the process, maybe to pilot something new. Personally, I would prefer to piss off 20% of users vs. 100% of them if possible. If you make changes and there is a problem discovered a few days after launching it, it might effect everything.
3. reverse engineering hundreds of attributes might be pretty challenging months or years down the road, even if you were the one who design it -- and it might be even harder for another user to pickup where you left off. A clever naming convention will be even more critical for your own sanity reasons, and certainly for the next guy. With that many fields, I would maybe put a prefix on groups of related attributes to make it more manageable within object manager. everywhere.
my suggestion:
design your monster process to be a master template, but build-in decisions that allow highly customized branches to only apply to certain groups of actual business processes. This way, you could make 5 copies of the master when it's ready, and apply 5 different windows to each process clone. to keep it organize, you could put letters on each process, and on each window title, and somewhere on each window.
Process v1 = master version 1, the only process you alter, but it's not used in production.
Process v1a = clone of process 1, cluster A (uses window A)
Process v1b = clone of process 1, cluster B (uses window B)
Process v1c = clone of process 1, cluster C (uses window C)
Process v1d = clone of process 1, cluster D (uses window D)
next time when you make changes, or add new processes:
Process v2 = master version 2, the only process you alter, but it's not used in production.
Process v2a = clone of process 2, cluster A (uses window A)
Process v2b = clone of process 2, cluster B (uses window B)
Process v2c = clone of process 2, cluster C (uses window C)
Process v2d = clone of process 2, cluster D (uses window D)
Process v2e = clone of process 2, cluster E (uses window E) for new stuff, but changes still made in master
That being said, I have never attempted anything so large, but I am very curious how you are making out.
Mike