Hi! Il 19/05/2013 16:09, Luca Chiodini ha scritto: > Big thanks to everyone who made possible CMS. We used it in a competition > among schools named ABC [0] and it worked very well even if it was running > on a low-specs machine. Nice! Any problem if we report this thing on our testimonials list[1]? [1] https://github.com/cms-dev/cms/blob/master/README.md > I want to report two facts: > a) CWS displayed wrong information about the execution time, when the > submission took more than 1 second ( in fact 1 second was the limit time). > I've noticed that this only happens when CMS is running on a VPS (i.e. this > doesn't happen when on a real machine); AFAIK it could be a problem on how > the sandbox tries to calculate the time. > This is a very annoying problem because a user sees that his submission > seems to need a very little time (e.g., 0.2 second) but is awarded 0 points > with a "Execution timed out" message. Bugs of this kind have already been reported. They depend on the grading sandbox and they should disappear with the new sandbox (which will be used starting from version 1.1). At least, so the voices that I heard say: I never really tried. > b) As stated on OII and IOI, "A submission that outputs the same results > regardless the input gets 0 point"'. But, despite this sentence, I haven't > seen any method to prevent this behaviour and I can't figure how it can be > easily resolved. I don't think anything like that was ever written in a IOI task description. It has been in use for a while in OII and I always strongly advised against it: actually, I've seen in many contexts where it was plainly false (outputting always the same result can bring you points, if that result happens to be right in at least a test case and there is no test case grouping). I never understood why this sentence happens to be so resistant to time: at some point I obtained to change it in some contests to "... gets no significant score", or something like that. But what do you exactly mean with "preventing this behavior"? This behavior actually doesn't exist: there are no provisions in CMS to forcibly score 0 a submission that proves independent from the input. Giovanni. -- Giovanni Mascellani <mascellani@xxxxxxxxxxxxxxxxxxxx> Pisa, Italy Web: http://poisson.phc.unipi.it/~mascellani Jabber: g.mascellani@xxxxxxxxxx / giovanni@xxxxxxxxxxxxxxxxxxxx
Attachment:
signature.asc
Description: OpenPGP digital signature
