emilgedda / hattis Goto Github PK

As topic, should be easy to fix.

For kattis problem with a point system, hattis reports testcases with an off by one (3 of 2) and submission rejected message even though it was accepted.

When having a compile error the total number of test cases are missing and the test case is off by one (1 of 0)

Perhaps some crude regex-solution?
Should be fairly easy, does not need to be 100% correct all the time.

If you are making quick edits and uploading to kattis, it can happen that you are timed out for certain time (usually 1 or 2 minutes).

Hattis receives a 404 error and the submission id becomes the seconds counter until one can submit again.

I propose a better description where it explicitly tells you that you've been timed out and that you can submit again in $duration_left seconds.

In problems which have point systems, it's very interesting to know how much time an algorithm took to execute. The number of iterations a Monte Carlo algorithm took to execute can be tweaked to fully utilize the given time for a problem to maximize a test score.

So after the line that reads accepted, a line printing the execution time / max allowed time could be printed.

By the way, I love this little bundle of love. Use it all the time, nice work @EmilGedda!

Display information such as memory limit and time limit from each problem.
Requires some new parsing code and a http request to the problem page.
(Async IO, forkIO/Mvars? promises?)

This should probably be displayed right after a submission fails or is accepted and not earlier and above compiler errors/test hints etc but below score and cpu time.

When submitting to a non-existent problem hattis returns Prelude.read: no parse.

This could have a more descriptive error message. This can happen since some challenges exists only on kth@kattis and are missing on open@kattis. Gave me a headache when I used the wrong kattisrc file.

And not write to files, in case the user has their own Makefile already in place.
Simply print the Makefile to stdout instead, and the user is free to redirect however they want.