I'm not sure how a file system concept corresponds to shared RDD or custom MapReduce implementation... May we should have another "assembly" just for in-memory file system?
-- Nikita Ivanov On Tue, Sep 29, 2015 at 7:04 PM, Dmitriy Setrakyan <dsetrak...@apache.org> wrote: > How about we rename the hadoop-accelerator to "in-memory file system" and > include spark and map-reduce there as well. > > "In-memory file system" is a known entity and will bring more interest. We > can position the map-reduce and spark RDD as additional features there. > > Thoughts? > > D. > > On Wed, Sep 30, 2015 at 4:00 AM, Nikita Ivanov <nivano...@gmail.com> > wrote: > > > Well, my thinking at this point is that none of these three modules would > > interest in real life. So, I was thinking of giving users a clear > direction > > on getting a) fabric only, b) hadoop accelerator, and c) spark > integration. > > > > -- > > Nikita Ivanov > > > > > > On Tue, Sep 29, 2015 at 5:29 PM, Konstantin Boudnik <c...@apache.org> > > wrote: > > > > > One of the reasons is that I don't believe that running two builds to > get > > > two > > > different module assemblies is a right or useful idea. I can use > reactor > > to > > > build just what's needed, but looks like it won't produce specific > > > assemblies, > > > because spark is included into core package. hence, I would have to > move > > > the > > > jars selection logic into the packaging, which seems orthogonal to the > > case > > > at hands. > > > > > > As for the intersection of the two: will Spark more often intersect > with > > > the > > > data fabric core, you think? > > > > > > Cos > > > > > > On Tue, Sep 29, 2015 at 10:58AM, Nikita Ivanov wrote: > > > > Cos - why mixing the two? I think if we are embark on this route - > why > > > not > > > > create two different assemblies, one for Hadoop bits, and another for > > > Spark > > > > integration? Two would rarely intersect IMHO. > > > > > > > > -- > > > > Nikita Ivanov > > > > > > > > > > > > On Mon, Sep 28, 2015 at 10:40 PM, Konstantin Boudnik <c...@apache.org > > > > > wrote: > > > > > > > > > While a minor modification which won't affect anything, I'd like > to > > > put > > > > > on the table the consolidation of hadoodp and spark bits into > > > > > ignite-accelerator assembly. The fix is ready and I was just > waiting > > > for > > > > > 1.4 to come out. > > > > > > > > > > Cos > > > > > > > > > > On September 28, 2015 11:11:44 AM PDT, Vladimir Ozerov < > > > > > voze...@gridgain.com> wrote: > > > > > >I'd like to add several things. > > > > > >1) It looks like fix for IBM JVM will brake backward compatibility > > > > > >because > > > > > >the problem is caused by missing serialVersionId, right? > > > > > >2) HP-UX users are also concerned, see last comments in > > > > > >https://issues.apache.org/jira/browse/IGNITE-1493 I think we must > > > > > >adress > > > > > >Unsafe refactoring in this release as well. > > > > > > > > > > > >On Mon, Sep 28, 2015 at 7:57 PM, Dmitriy Setrakyan > > > > > ><dsetrak...@apache.org> > > > > > >wrote: > > > > > > > > > > > >> Gianfranco, > > > > > >> > > > > > >> I am not too familiar with JBPM. Can you make a case on why we > may > > > > > >need > > > > > >> this integration and how it will benefit Ignite users? > > > > > >> > > > > > >> Thanks, > > > > > >> D. > > > > > >> > > > > > >> On Mon, Sep 28, 2015 at 6:56 PM, Gianfranco Murador > > > > > ><mura...@apache.org> > > > > > >> wrote: > > > > > >> > > > > > >> > I would like to propose the integration with JBPM , > > > > > >> > I think I can finish in a couple of weeks since I have a bit > of > > > > > >time to > > > > > >> > devote to Ignite. > > > > > >> > Let me know if you agree. > > > > > >> > Regards, Gianfranco > > > > > >> > > > > > > >> > 2015-09-28 18:50 GMT+02:00 Dmitriy Setrakyan > > > > > ><dsetrak...@apache.org>: > > > > > >> > > > > > > >> > > I would like to add one more: > > > > > >> > > 10. OSGI integration - Raul Kripalani > > > > > >> > > > > > > > >> > > Raul, hopefully you will have time to finish it. > > > > > >> > > > > > > > >> > > Also, if anyone in the community wants to propose a feature > > they > > > > > >are > > > > > >> > > working on which can be finished within next couple of > weeks, > > it > > > > > >should > > > > > >> > get > > > > > >> > > into 1.5 as well. > > > > > >> > > > > > > > >> > > D. > > > > > >> > > > > > > > >> > > On Mon, Sep 28, 2015 at 6:43 PM, Yakov Zhdanov > > > > > ><yzhda...@apache.org> > > > > > >> > > wrote: > > > > > >> > > > > > > > >> > > > Guys! > > > > > >> > > > > > > > > >> > > > We have finalized 1.4 release! > > > > > >> > > > > > > > > >> > > > Thanks to everyone involved! That was very important > release > > > > > >for our > > > > > >> > > > project. > > > > > >> > > > > > > > > >> > > > Let's discuss scope for 1.5. > > > > > >> > > > > > > > > >> > > > I think the main focus areas should be: > > > > > >> > > > 1. Optimized marshaller rework - Alex Goncharuk works on > > this. > > > > > >> > > > 2. .Net and Cpp API release (depends on 3) - Vladimir > Ozerov > > > > > >> > > > 3. Portable API release - Denis Magda and Vladimir > > > > > >> > > > 4. Failover for continuous query notifications - Nick > > Tikhonov > > > > > >> > > > 5. Cache Rebalancing optimizations - Anton Vinogradov > > > > > >> > > > 6. Fix compatibility with IBM JDK > > > > > >> > > > 7. Cache stability and performance fixes (as many as > > possible) > > > > > >> > > > > > > > > >> > > > Nice to have: > > > > > >> > > > 8. Finish vert.x integration - Andrey Gura works on this. > > > > > >> > > > 9. Finish node.js integration - I think Semyon should pass > > it > > > > > >over to > > > > > >> > > > someone else, given he is busy with cache. Anyone? Andrey > > > Gura? > > > > > >> > > > > > > > > >> > > > Comments are welcome! > > > > > >> > > > > > > > > >> > > > [1] https://issues.apache.org/jira/browse/IGNITE-950 > > > > > >> > > > [2] https://issues.apache.org/jira/browse/IGNITE-1282 > > > > > >> > > > [3] https://issues.apache.org/jira/browse/IGNITE-1258 > > > > > >> > > > [4] https://issues.apache.org/jira/browse/IGNITE-426 > > > > > >> > > > [5] https://issues.apache.org/jira/browse/IGNITE-1093 > > > > > >> > > > [5] https://issues.apache.org/jira/browse/IGNITE-324 > > > > > >> > > > [6] https://issues.apache.org/jira/browse/IGNITE-1526 > > > > > >> > > > [7] https://issues.apache.org/jira/browse/IGNITE-1345 > > > > > >> > > > [8] https://issues.apache.org/jira/browse/IGNITE-1079 > > > > > >> > > > [9] https://issues.apache.org/jira/browse/IGNITE-961 > > > > > >> > > > > > > > > >> > > > --Yakov > > > > > >> > > > > > > > > >> > > > > > > > >> > > > > > > >> > > > > > > >> > > > > > > >> > -- > > > > > >> > Gianfranco Murador > > > > > >> > Igniter and Software Engineer. > > > > > >> > > > > > > >> > > > > > > > > > > >
