Specifying a filespec that only looks at the main codeline, doing a
'p4 changes -i ...' or 'p4 fixes -i ...' seems to ignore completely
whether or not the full changelist has been 'merged in'. As long
as one file has been 'merged', then these commands report positively.
Yes, that's how it works, and the reason we chose not to rely on -i at
Network Appliance.
We're currently machinery external to Perforce to track change
propagations between codelines. It involves out defect tracking
integration, and a database that is, logically, a replacement for the
"job" and "fix" records maintained by Perforce.