Thanks to everyone for highlighting the issues with Guava. As Jena is a
library much of the time, gettign tangled in
I've just found out that jena-csv uses Guava already.
Given Elephas, HADOOP-10101 is a bit of an issue.
Shading is certainly a possibility (or, worse, like
com.sun.org.apache.xerces).
The immediate caching need is for TDB so a possible way forward is to
have the dependency for TDB, and TDB contain the caching implementation
of it's own choice for the LRU cache. This assumes TDB and Hadoop
aren't in the same JVM.
Andy
