Home | About | Sematext search-lucene.com search-hadoop.com
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB
 Search Hadoop and all its subprojects:

Switch to Threaded View
Kafka >> mail # dev >> Re: Kafka and Scala versions


Copy link to this message
-
Re: Kafka and Scala versions
Ok. Thanks. We will look into upgrading to 2.8.2 post 0.8.

Jun
On Mon, Apr 8, 2013 at 2:08 PM, Scott Carey <[EMAIL PROTECTED]> wrote:

> We are on 2.10 already.
>
> String interpolation (2.10), Value classes (2.10), parallel collections
> (2.9), are too good to give up.  Akka actors, futures, and promises being
> part of the distribution are nice as well.
>
> The compiler is faster, and emits a lot of warnings in Kafka code that
> could be bugs in Kafka.
>
> Upgrading to 2.8.2 significantly simplifies the ability to cross-compile
> with 2.10 (only one small file needs to have a version that differs,
> instead of ~1500 lines).  See my recent patches in KAFKA-717.
>
>
> On 4/7/13 10:17 PM, "Jun Rao" <[EMAIL PROTECTED]> wrote:
>
> >Scott,
> >
> >Yes, we are still using scala 2.8.0 at LinkedIn. We do plan to upgrade to
> >a
> >later version of scala at some point. However, this may take some time
> >since we will need to upgrade other scala projects at LinkedIn at the same
> >time.
> >
> >It's unfortunate that code change is needed in order to support scala
> >2.10.
> >What are the things that you are looking for in scala 2.10?
> >
> >Thanks,
> >
> >Jun
> >
> >
> >On Fri, Apr 5, 2013 at 1:10 PM, Scott Carey <[EMAIL PROTECTED]>
> >wrote:
> >
> >> Kafka 0.8 and trunk by default have SBT set up to cross-compile 2.8.0,
> >> 2.8.2, 2.9.1 and 2.9.2, and it defaults to 2.8.0.
> >>
> >> I assume that one of the major contributors requires 2.8.0.  Any
> >> approximation on how long will that last?
> >> Dropping 2.8.0 would be beneficial because some uses of Java Conversions
> >> is replaced with Java Converters (in 2.10 +), and Java Converters was
> >>not
> >> added until 2.8.1.
> >>
> >>
> http://stackoverflow.com/questions/8301947/what-is-the-difference-between
> >>-j
> >> avaconverters-and-javaconversions-in-scala
> >>
> >> https://issues.apache.org/jira/browse/KAFKA-717 is attempting to add
> >>build
> >> support for Scala 2.10.  The required changes to support 2.10 use Java
> >> Converters, which is not available in 2.8.0.
> >>
> >> The other item required to support Scala 2.10 is to move from
> >> scala.StaticAnnotation to scala.annotation.StaticAnnotation.
> >> Scala 2.10.x has only scala.annotation.StaticAnnotation.
> >> Scala 2.9.x has both scala.StaticAnnotation (deprecated) and
> >> scala.annotation.StaticAnnotation.
> >> Scala 2.8.x has only scala.StaticAnnotation
> >>
> >> Dropping support for 2.8.x entirely and supporting only 2.9.x and 2.10.x
> >> would make 2.10.x support easy.
> >> If 2.8.2 is required (but not 2.8.0), then some sort of conditional
> >> compile or two flavors of the source code where
> >> scala.annotation.StaticAnnotation is used will be required.  Use of
> >> StaticAnnotation is very limited.
> >> If 2.8.0 must be kept, then two versions of all code that uses Java
> >> Conversions APIs will also need to be maintained.
> >>
> >> I can't find any discussions about 2.8.0 usage requirements.  I also
> >>don't
> >> know enough about sbt to configure cross compilation to use different
> >> source files for different cases, which would solve the problems above
> >>but
> >> add some maintenance burden.
> >>
> >>
> >>
>
>

 
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB