I am sorry it took a while to get back to this. For http caches we support clearing of the caches within a date range (the last accessed date) and for a subset of chosen URLs. Date range is easy to check for code caches as well. For the URL, since we use two keys we need to derive the URL. Currently the generated key is 
_key+resource_url+"  \n"+origin. I plan to extract the resource_url and then check if it matches the URLs that needs to be cleared. Since the separator (" \n") cannot be in a valid URL it would be safe to do this. This also means that I clear the entries from all origins corresponding to the URL. 

kinuko@, morlovich@ does this sound reasonable?