Hello, Today Kylin's binary package ships an Apache Spark binary, the total package is 297 MB in v2.5.0; I noticed that if the package is above 300 MB, it would be rejected by Apache SVN. Now in the whole package, Spark takes about 200 MB. Later when we upgrade to a higher version, the package can be bigger. Without spark, the package is around 100 MB.
Since we don't have customization to Spark, we can provide a download-spark.sh script for the user to easily download. If the user already has that in local, he just needs to copy or create a symbolic link in Kylin folder. This will reduce 2/3 package size, and save time/bandwidth for distributing Kylin. If you see a problem this change may bring, please complain. -- Best regards, Shaofeng Shi 史少锋
