Can compiled CTF files be retrieved from HDFS?

Following the instructions to run on Hadoop, I've got the basic airlinesmall sample running. I was able to use `hdfs:///` paths for the input file `hdfs:///airlinsmall.csv` and output folder `hdfs:///airlinesmall-output`. However, I had to put the `ctf` file on the filesystem of the master VM. I really want to be able to use `hdfs:///airlinesmall.ctf` as the location for the input file instead of `/home/cameron/airlinesmall.ctf`. Am I doing something wrong? If it is not supported, could this be added?

I'm using R2015b (Runtime v90). <file:///usr/local/MATLAB/MATLAB_Runtime/v90/toolbox/mlhadoop/jar/a2.2.0/mwmapreduce.jar> com.mathworks.hadoop.MWMapReduceDriver

It if funny that the first lines from the output show it being uploaded to HDFS. Couldn't that be skipped and set the HDFSCTFPATH directly?

java.library.path: /usr/lib/hadoop/lib/native
HDFSCTFPath=hdfs://cluster-for-cameron-m:8020/user/root/airlinesmall/airlinesmall.ctf
Uploading CTF into distributed cache completed.

[Nikhil Sreekumar]

In hadoop, usually in deployable mode, the jar file is deployed after compiling. In MATLAB, ctf files are analogous to jar files, hence they are loaded into the distributed cache. If you find any link mentioning copying of the executable to the hdfs, could you share?