Did you just try it? That's the best thing to start with in your own environment.
Next when the results do not match your expectation please do ask here making clear what you did, what you expected, what you saw, the replicate version/platform, the source and target DB's with version if you can).
On SQLserver you wouldn't get very far getting:
--- Cannot drop the table 'dbo.test' because it is being used for replication.
On Oracle the table object ID will change and the running task is only looking for the old ID and will not see the new change.
Now without thinking about it too much about it I expected I just needed a stop+resume, for a table dropped and recreated with ignore drop on CDC in the task. I was a little surprised that did NOT pick up the changes in a crude test. To see new changes I had to re-load loosing the old contents which the CDC 'ignore drop' had protected. I suppose I should have change my full-load setting to 'do nothing' to keep the old data if needed. Hmmm. maybe I'll try some more after a beer or two, but not now. You try!
Reload will get everything working, which is simple and safe. It may be more productive to work on the mission definition and goals than the technology: Why/Whed are tables on source re-created. What is the desired action on the target? Should/Could the replication be stopped when this is happening? Those are more business questions than Replicate question within the constraints of what Replicate can do.
hope this helps some,
Hein.
