I am hitting the same question as you.
When using streaming replication, it is in theory not needed to ship the archived Xlogs to the standby.
But the archived Xlogs are also needed for normal recovery and they can also be needed if the standby cannot catch up with the master (for example if it was stopped for a while). Note that I am not 100% sure that archives are needed in this case because there is also the concept of replication slot by which the master is aware of what has been consumed by the slave (my understanding). With replication slots I think the master will accumulate the XLOGS as long as they are needed by the slave (and stop operating if the slave does not come back and the file system is full ?)
What I will do is ship them to the slave and even keep a copy on the master.
So, on the master
archive_command = rsync -ac %p postgres@other-pg-server:/u01/archive/%f
However, I think I will use a shell script that will copy the file locally and rsync it to the slave. This shell script can also check if it is running on the slave and if yes do nothing (this way the config file can be symetric between the two).
On the slave, in recovery.config,
restore_command = 'cp /u02/archive/%f %p'
I am using a replication slot but I will still need to do the cleanup of the archived Xlogs, via a cron jobs that delete files older than x days. I did not investigate the pg_archivecleanup program yet (https://www.postgresql.org/docs/9.2/static/pgarchivecleanup.html)