Marco, 945 is a bit simpler. I would also start with it. What you need to do is to replace transaction based logic inside GridCacheCommandHandler#appendOrPrepend() with org.apache.ignite.IgniteCache#invoke() preserving the semantics. If we remove explicit transaction start and change the whole thing to a single transform we have logic that works for TX and ATOMIC caches and produce the same results.
I also think you can remove the logic which submits cache operation to a separate thread because it is not needed any more - ctx.closure().callLocalSafe(..). You should return future for invoke operation. That will be enough. I filed another ticket to revise other handlers for this shortcoming ( https://issues.apache.org/jira/browse/IGNITE-957). As far, as LOG handler - you are right http handler for it is missing, but now I am thinking whether we still need it. It seems we can skip it and remove the command from API. Marco, you should start implementation in a branch made out from ignite-sprint-5 and then attach patch to JIRA issue. Please also feel free to assign it to yourself (IGNITE-945). Let me know if you have more questions. --Yakov 2015-05-28 21:30 GMT+03:00 Marko Jevtic <markojev...@gmail.com>: > I’d take IGNITE-945, but I would need some pointers :) > > I looked at master branch, there’s no handler for LOG rest command, … > unless LOG command should be registered in some other way I’m not seeing… > > I assume this is the description of the command in case handler should be > implemented: > https://apacheignite.readme.io/v1.0/docs/rest-api#log < > https://apacheignite.readme.io/v1.0/docs/rest-api#log> > > Should I be looking at the sprint-5 branch only ? > > > > > On May 28, 2015, at 5:29 PM, Yakov Zhdanov <yzhda...@apache.org> wrote: > > > > Guys, > > > > Tickets are waiting for their contributors! > > > > https://issues.apache.org/jira/browse/IGNITE-944 > > https://issues.apache.org/jira/browse/IGNITE-945 > > > > --Yakov > >
