I've compiled my code against 2.11.8 and still couldn't import via magic with the same error. Neither toree_opts worked, but the solution from here did:
https://www.mail-archive.com/dev@toree.incubator.apache.org/msg01649.html --spark_opts="--jars=/usr/tmp/jars/CustomJar.jar" On Mon, May 21, 2018 at 3:50 AM, Marius van Niekerk < marius.v.niek...@gmail.com> wrote: > Making toree work under scala 2.12 will require recompiling toree as well > as probably some changes to the interpreter code. > > On Sat, May 19, 2018, 08:48 Igor Fialko <igorfia...@gmail.com> wrote: > > > Thanks, I'm using toree 0.2-dev prebuilt. I'll try to compile myself > while > > keeping the scala version same across the parts and report back. > > > > Could you please confirm that I'm using the 'toree_opts' correctly, > because > > they don't seem to be parsed. (No debug-level output and no message about > > that the jar in path was loaded or not) > > > > sudo jupyter toree install --spark_home=/path/bin/spark-2.3.0 > > --toree_opts='--debug --jar-dir /path/target/scala-2.12' > > > > On Sat, May 19, 2018 at 3:32 AM, Luciano Resende <luckbr1...@gmail.com> > > wrote: > > > > > Current master, which is the one supporting Spark 2.x is currently > > compiled > > > with Scala 2.11. Also, if you are using Scala 2.12, you should make > sure > > > that both Spark and Toree gets compiled properly with Scala 2.12 as I > > > believe Spark still provide binaries compiled with Scala 2.11. > > > > > > On Fri, May 18, 2018 at 3:39 PM, Igor Fialko <igorfia...@gmail.com> > > wrote: > > > > > > > Hello community, > > > > > > > > name is Igor and I'm working in my spare time on a trading system. So > > > far I > > > > used jupyter notebook with python kernel for visualisation/playing > > around > > > > with data. I wanted to go a step forward and reuse my existing scala > > > code, > > > > so I came up with toree as seemingly the only one still maintained > > > jupyter > > > > kernel. > > > > > > > > Therefore I've assembled a jar via sbt and tried to add it in the > > > notebook. > > > > It has failed with an exact same error as here. > > > > > > > > https://stackoverflow.com/questions/46403426/adding- > > > > dependency-to-apache-toree-fails-on-import > > > > > > > > 2018-05-18 23:43:51 WARN MagicManager:111 - Magic AddJar failed to > > > execute > > > > with error: > > > > object Predef does not have a member classOf > > > > scala.reflect.internal.FatalError: object Predef does not have a > member > > > > classOf > > > > > > > > > > > > 2018-05-18 23:43:51 ERROR OneForOneStrategy:159 - object Predef does > > not > > > > have a member classOf > > > > scala.reflect.internal.FatalError: object Predef does not have a > member > > > > classOf > > > > > > > > So I've tried to import the jar via toree_opts like > > > > > > > > sudo jupyter toree install --spark_home=/path/bin/spark-2.3.0 > > > > --toree_opts='--jar-dir /path/target/scala-2.12' > > > > > > > > which didn't make any impact unfortunately (from source code I could > > see > > > > that it should print a message that jar could not be loaded, neither > > did > > > an > > > > import work). > > > > > > > > I could assume that it's maybe a scala version issue? My project is > > > > compiled against 2.12.3, I couldn't find scala version that toree is > > > using. > > > > > > > > Did somebody face this issue? How could I debug what is wrong? > > > > > > > > Thanks > > > > Best regards, > > > > Igor > > > > > > > > > > > > > > > > -- > > > Luciano Resende > > > http://twitter.com/lresende1975 > > > http://lresende.blogspot.com/ > > > > > > -- > regards > Marius van Niekerk >