Obviously this patch also neglected to add support to the C client (though honestly that's a lot harder than the java side). If you don't plan to do the C client, then I can pick up that work.
On Thu, Mar 8, 2012 at 5:10 PM, Ted Dunning <[email protected]> wrote: > Looks like a cut and paste error in the first line of the GetData > constructor: > > + super(ZooDefs.OpCode.setData, path); > > > Perhaps you meant ZooDefs.OpCode.getData here? > > The check for permissions also seems wrong here: > > > + checkACL(zks, nodeRecord.acl, ZooDefs.Perms.WRITE, > request.authInfo); > > > As far as really reviewing this, it would really help to have it in > reviewboard instead of just reading a diff. > > Also, this only seems to handle getData and doesn't seem to include any > tests. Is the omission of getChildren intentional? Do you plan to have > tests? > > On Thu, Mar 8, 2012 at 3:54 PM, Ted Yu <[email protected]> wrote: > > > Please take a look at the patch I attached to ZOOKEEPER-1407. > > > > Cheers > > > > On Thu, Mar 8, 2012 at 2:52 PM, Ted Dunning <[email protected]> > wrote: > > > > > Yes. > > > > > > It should be straightforward. There are just a number of places to > > touch. > > > You need a factory in Op, a sub-class to hold the transaction, then on > > the > > > server side there are 2-4 switch statements that need to be inspected. > > > Marshal can comment whether the commit code needs change, but I would > > > expect it. > > > > > > On Thu, Mar 8, 2012 at 2:39 PM, Ted Yu <[email protected]> wrote: > > > > > > > There are GetDataRequest / GetDataResponse in src/zookeeper.jute > > > > > > > > Would the new GetDataTxn be able to reuse them ? > > > > > > > > Thanks > > > > > > > > On Wed, Mar 7, 2012 at 8:06 AM, Ted Yu <[email protected]> wrote: > > > > > > > > > https://issues.apache.org/jira/browse/ZOOKEEPER-1407 has been > > logged. > > > > > > > > > > > > > > > On Tue, Mar 6, 2012 at 10:41 PM, Ted Dunning < > [email protected] > > > > >wrote: > > > > > > > > > >> The use cases that I came up with for multi all involved reads > done > > > > before > > > > >> multiple writes as with a read-modify-write operation. In that > > case, > > > > the > > > > >> reads need to be done before the updates. Those use cases > motivated > > > the > > > > >> Check operation so that you can ensure that nothing has changed > > before > > > > the > > > > >> updates are done. > > > > >> > > > > >> But as Marshal says, there isn't any reason not to have them. > > > > >> > > > > >> On Tue, Mar 6, 2012 at 10:10 PM, Ted Yu <[email protected]> > > wrote: > > > > >> > > > > >> > On first glance, Op doesn't seem to cover getData(). > > > > >> > > > > > >> > Is this intentional ? > > > > >> > > > > > >> > How do I read the data for multiple znodes in one transaction ? > > > > >> > > > > > >> > Thanks > > > > >> > > > > > >> > > > > > > > > > > > > > > > > > > > >
