Hi Taher, Maybe we should discuss this in another thread.[1]
In my opinion, it is valuable to integrate with Beam and with Flink. In fact, as far as I know, Beam and its Spark and Flink runners are not widely used, at least in China. There are still a large number of users, either using Spark or using Flink. [1]: https://lists.apache.org/api/source.lua/1533de2d4cd4243fa9e8f8bf057ffd02f2ac0bec7c7539d8f72166ea@%3Cdev.hudi.apache.org%3E Best, Vino taher koitawala <[email protected]> 于2019年7月31日周三 上午2:07写道： > I think lastly we left on this discussion is when we questioned if we > should port the whole of hudi to beam and then use Flink or Spark runner > there instead of writing hudi separately for Flink is what I recall. > > > > On Tue, Jul 30, 2019, 11:14 PM Vinoth Chandar <[email protected]> wrote: > > > Hi, > > > > Could not agree more. Its captured under the work for the first release > > already https://issues.apache.org/jira/browse/HUDI-121?filter=-1 > > > > Balaji is the RM. Plan to do this in August. > > > > One issue we realized was that we need a solid migration path, since the > > tables are all registered with com.uber.hoodie.HoodieInputFormat as the > > input format. > > We plan to do a HIP around this, once we get past the jar/bundle > redoing.. > > (thats another logical step we are considering before doing this, to test > > with older setups more easily) > > > > Hope that helps > > /thanks/vinoth > > > > On Mon, Jul 29, 2019 at 11:20 PM vino yang <[email protected]> > wrote: > > > > > Hi guys, > > > > > > I find hudi still uses the package pattern like "com/uber/hoodie". > > > > > > Since it has joined the ASF incubator, should it follow the Apache > > package > > > naming rules? For example: org/apache/hoodie(hudi)? > > > > > > Is there any plan about renaming? > > > > > > Best, > > > Vino > > > > > >
