I think this sounds very reasonable, +1 to the schedule.
I would definitely add
- FLIP-10 (unify checkpoints and savepoints)
- FLIP-7 (metrics in web UI)
- FLIP-12 (async request operators)
Those should be all safe bets, as they are basically done.
Let's see what else is in shape until the feature freeze day.
On Mon, Oct 17, 2016 at 2:18 PM, Fabian Hueske <fhue...@gmail.com> wrote:
> Hi everybody,
> Flink 1.1.0 was released in August and I think it is time to kick off a
> discussion about the schedule and scope of Flink 1.2.0.
> == Scope
> We started to collect features for Flink 1.2.0 in the Flink Release wiki
> page .
> I copy the feature list for convenience:
> - Dynamic Scaling / Key Groups (FLINK-3755
> <https://issues.apache.org/jira/browse/FLINK-3755>) (Design Doc
> - Rescalable Non-Partitioned State (FLIP-8
> - Backwards compatible savepoints (FLINK-4797
> Integrate Flink with Apache Mesos (FLINK-1984
> <https://issues.apache.org/jira/browse/FLINK-1984>) (Design Doc
> - Secure Data Access (FLINK-3930
> <https://issues.apache.org/jira/browse/FLINK-3930>) (Design Doc
> - Queryable State (FLINK-3779
> - Metrics in Webinterface (FLIP-7
> - Kafka 0.10 support (FLINK-4035
> - Table API: Group Window Aggregates (FLINK-4691, FLIP-11)
> - Table API: Scalar Functions (FLINK-3097)
> Is a feature missing or something too ambitious?
> == Schedule
> In my opinion we should try to release 1.2.0 before Christmas. That means,
> we should have a RC1 by end November / early December and a feature freeze
> two weeks earlier in mid of November. Does that sound appropriate or should
> we aim for an earlier or later release?
> What do others think about the scope and schedule?
> Please discuss.