Yavuz, I agree adding more options is not in the best interest of the users. +1 to the idea of looking at what is in the Spark distribution. Will update the jira description to reflect that idea. On Mar 18, 2016 5:15 PM, "Yavuz Nuzumlali (JIRA)" <j...@apache.org> wrote:
> > [ > https://issues.apache.org/jira/browse/TOREE-280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15202266#comment-15202266 > ] > > Yavuz Nuzumlali commented on TOREE-280: > --------------------------------------- > > What about the following option: > > - Since we know SPARK_HOME (either given by user or the default value), we > could look inside the ??$SPARK_HOME/python/lib?? folder for a file name > matching py4j, and add that as the value. > > I think holding a mapping would add a level of complexity which seems > redundant. > > Also, other alternatives does not seem to be convenient for users, adding > SPARK_HOME and the interpreter should be enough for toree to run. > > > PYTHON_PATH Needs To Be Configured At Install > > --------------------------------------------- > > > > Key: TOREE-280 > > URL: https://issues.apache.org/jira/browse/TOREE-280 > > Project: TOREE > > Issue Type: Improvement > > Reporter: Corey A Stubbs > > Priority: Minor > > > > When installing the interpreters, the PYTHON_PATH env var will need to > change based on the spark distribution. There are a couple paths to do this. > > 1. We can allow the PY4J version to be passed > > 2. Figure out the SPARK_VERSION and keep a mapping of the PY4J versions > and substitute the value > > 3. The user must specify the value for the PYTHON_PATH > > > > -- > This message was sent by Atlassian JIRA > (v6.3.4#6332) >
