enable connection in cluster mode for output in client machine

While working on Spark I implemented a feature where when you submit a Spark job you enable the option that when it finishes the client machine will read from a file that the Spark job wrote to. It enables a quick and easy way to get output of lets say a dynamic Spark SQL query in the client without manually doing a hdfs download command or anything like. There are some pre requisites for it ( like having configurations for the hdfs communication) but I think its completely possible to  implement it and help many users to have an established communication line with a Spark job.

What do you think about the option? 
Would be glad to hear any issues with this idea or why it shouldnt be implemented.