The current test suite requires adding the programs to db-test.js file. This can grow up to become big, monolithic and janky. We should find a better solution to arrange the tests.

They should go in languages folder. Take inspiration from C.js.

• CPP
• Java
• Python

Give suggestions for more languages in the comments.

Followed the steps given in the wiki to run this judge. Did everything, but it is not working on the localhost, not on any port. The page is just loading when tried to enter localhost followed by port number.
There is no output in the browser except it is loading forever. And only the terminal says "Slave server is up" and "Master server is up".

The programs submitted by users might have infinite loops or might take too much time to run, so we should have something that allocates a fixed time period, a time box, to each code run.

We need this to ensure that all child processes are killed by the sandbox.

when i run your code in my system. it show many bugs and a few package or plugin like fakechroot isnt support windows

Who needs breakages that happened in Codejudge v1?

We need this test to assert that out judge successfully works in high load conditions. We had something similar for manual testing in previous commits.

Now that we have automated testing we should add it back appropriately. The test should do something like open 100 parallel connections and check that the judge works as expected.

This is a purely performance reason. This article states that for...in loops are not optimized by V8.

Need a good failure recovery mechanism when a slave is dead. Code should go here.

The judge is useless without this.

There is no security sandbox for the code run right now. Add them here.

The db-handler should be able to specify the memory limit.

To check if we kill the program for high memory usage.

Take a look at ochko/safeexec for a better sandbox and implement it here.

Right now the promise chain in slave.js reports error at every step when a promise is rejected. This leads to the same error output multiple times. We should change to so that the reject handler is called only once.

This was detected by @rohilsurana when working on #20.