M2 Assignment Feedback
I have two pieces of feedback for the M2 milestone so far that I hope can be addressed by M3:
1. The starter code is out of date and contains irrelevant material. I'm of the opinion that providing starter code with an interface that is essentially 100% inconsistent with the handout isn't that helpful and adds a ton of confusion.
2. The autotester is massively unstable. I don't think it's acceptable to have to wait for manual intervention everytime something breaks with the system.
It seems that the filesystem is not properly cleared between test runs, leading to merge conflicts when rewriting git history (https://piazza.com/class/lctgp0rn94237x/post/84). Additionally, the docker containers running the services do not get killed when the tester restarts, leaving to situations like this (https://piazza.com/class/lctgp0rn94237x/post/146). It was theorized that this error is a side effect of queuing multiple tests concurrently -- but this really seems like a flaw with the tester. Queueing a new job should not blindly make it run without killing off the previous test.