Improvements

[macfreek]

Hello Hadi,

I just took some time to isolate the public improvements from the assignments, and pushed them to a fork. Feel free to merge or cherry-pick whatever you think is useful.

Commit	Change
47f967a	Properly delete output file from HDFS. If it is not deleted, a second run of the program raises a Java exception.
a9b10d2	Ignore internal_use/tmp directory
4b860bf	Remove Python -u option. This did not work on Hortonworks VM for me.
365d5ae	Use modern toolchain: hdfs and yarn instead of hadoop
bbeddcc	Don’t hardcode paths in Java code. This failed on a local Hadoop cluster I have access to.
d9611bc	Workarounds to work on Mac OS X: (1) Use md5 (from Openssl) instead of md5sum (from GNU coreutils) is unavailable. (2) Enforce Java 1.7 output (To be more correct, I should also set the bootstrap class path).
0aae948	Improved run.sh script: (1) Check if settings.sh is sourced. (2) -v option (verbose) show commands as they are executed. (3) Allow selective execution of assignments. E.g. run.sh A D to run assignments A and D. The default is to run them all.

Regards,
Freek

PS: it is shocking to see how many people push their assignment solutions to a public repository, thus violating the third clause of the Coursera Honour Code!

PS2: I just realized that none of these commits have a sign-off line. In case you appreciate it: Signed-off-by: [email protected]. A.k.a.: yeah, I really am to blame me for any bug introduced by these commits ;).

[xldrx]

Thanks Freek,
Nice work.
I will merge these issues in the master as soon as course is ended.