Hi Juan Carlos!

Thanks for the fast reply! It turned out to be very helpful, since you've forced me to prepare a reproducible example. So I tried to come up with a python sample and figured out that it worked. Then I refactored the actual Scala sample and saw that the problem was not the network but some misleading "java.lang.OutOfMemoryError: unable to create new native thread", which doesn't indicate a memory issue but has to do with getting new threads from the operating system. As a consequence I increased the number of processes in the jail config file, but without any effect. More or less by accident, I figured out that one can set the number of processes per submission in the VPL-Moodle plugin. After increasing the number of processes there from 16 to 40, everything worked!

So I can only repeat myself and thank you for this great tool!

Here's my currently used scala_run script that we use instead of the default one:

. common_script.sh
check_program scala
check_program scalac
APP=${VPL_SUBFILE0%.*}
scalac $VPL_SUBFILE0
cat common_script.sh > vpl_execution
echo "scala -nocompdaemon -classpath ./ $APP" >> vpl_execution
chmod +x vpl_execution

Compared to your current default script it assumes that the scala code is an application like follows:

object MyApp extends App {
...
}

This is way more common than writing scripts. The file, however, has to have the same name as the object defined in it (in our case "MyApp.scala", which is the same as in Java). Then this file is compiled and the app started. Probably you might use this as a new default script!

Regards,

Peter