On Thu, Dec 17, 2015 at 4:33 PM, Han Zhou <zhou...@gmail.com> wrote: > Hi Andy, > > Thanks for your review. > > > On Thu, Dec 17, 2015 at 2:00 PM, Andy Zhou <az...@ovn.org> wrote: > >> >> >> On Tue, Dec 15, 2015 at 7:06 PM, Han Zhou <zhou...@gmail.com> wrote: >> >>> Cached json objects were reused when sending notifications to >>> clients. This created a problem when there were different versions >>> of monitors coexiting. E.g. clients expecting version2 notification >>> would receive messages with method == "update2" but payload in >>> version1 format, which end up failure of processing the updates. >>> >>> This patch fixes the issue by using dedicated cache for each version. >>> >>> >> Is it really necessary to have json cache array? How about extending >> ovsdb_monitor_json_cache_node to also include version, than multiples >> versions >> of json_object can co-exist in a single cache. What do you think? >> >> > Well this could be an alternative solution. Since dbmon doesn't know which > versions of objects will be required later at the moment when the cache > node is being constructed, what we can do is to compose the object only for > the version of the first jsonrpc_monitor, and when a new version of > jsonrpc_monitor comes, compose for the new version. > That means we will need to redefine the search function as "search and > then check the version in the node", and redefine the insert function as > "search and then update for new version (if hash node not found, create new > node)". > > This should work but I am not sure if there is strong benefit compare to > the current patch which seems to be more straightforward. What do you think? > > Sorry that I misunderstood your comment. I thought in a more complex way :( I will provide v2 based on your suggestion. Thanks!
Han -- Best regards, Han _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
