- DOMjudge-devel - domjudge.org

DOMjudge hackathon(s)
by Tobias Werth 21 Aug '17

21 Aug '17

Hi, we want to do one or two hackathons in ~August to make progress on the implementation of the new REST API and event feed. See https://clics.ecs.baylor.edu/index.php/Main_Page and http://hera.ecs.csus.edu/pipermail/cliccs/ for the current state and discussions. Please fill out the following doodle if you want to participate: https://doodle.com/poll/dchzfbyci32esgh4 Tobi

2 3

Using the integrated event feed
by Phúc Khang Nguyễn 09 Aug '17

09 Aug '17

Hi, I'm currently running DOMjudge v5.1.3. Recently, I wanted to use the Resolver in the ICPCTools. I saw the integrated event feed in /www/feed but I really don't know how to use and I can't find the documentation of the event feed. I'm not a PHP user so I can't come up with any solution on my own. Could you show me how to set it up in order to use the Resolver or link me to the documentation? Best Regards, Khang *Nguyễn Phúc Khang* Mobile: 0931 633 723 Trường Đại học Khoa Học Tự Nhiên, ĐHQG-HCM

2 1

HELP - Writing DOMjudge on to hard disk
by Kevin Paul 13 Jul '17

13 Jul '17

I'm trying to install DOMjudge live on my hard disk as a partition. But I'm new to this and there isn't a lot of information on how to do this. Please help. Thanks in advance.

2 2

Out of range value for column 'totaltime'
by Richard Lobb 12 Jul '17

12 Jul '17

Hi guys We're running domjudge 5.1.1 and for some reason I don't understand we're being plagued this year by errors like the following: error: Error while executing curl POST to url http://domserver.cosc.canterbury.ac.nz/api/judging_runs: http status code: 500, response: error: SQL error, Error#1264: Out of range value for column 'totaltime' at row 1, query: 'REPLACE INTO scorecache_public (cid, teamid, probid, submissions, pending, totaltime, is_correct) VALUES (24,1403,222,1,0,-15780,1)' They occur when you set up a contest with an early opening date for testing purposes, make submissions then, later, move the opening date forward to the correct time. Refreshing the cache then results in those errors (because the submission time relative to the contest start time is negative), crashing the judgehosts. I've been solving the problem by marking the early submissions as "IGNORE this submission" or by hacking at the database directly but neither of those workarounds is very satisfactory. I don't recall having these errors last year, and nor do I think I've upgraded domjudge system since then, though I might have upgraded the OS? Any suggestions for a better work around, please? Richard

2 2

Checktestdata as EDSL
by Jeroen Bransen 30 May '17

30 May '17

Hi all, Checktestdata is a nice DSL and has proved helpful in many situation, but from the moment I used it I was missing features and wondering why it wasn't implemented as an EDSL (Embedded Domain Specific Language). Now, many years later, I've finally found some time to create it. I've implemented a little library in Haskell, and as a sneak preview I've taken the interception problem from NCPC 2016 as referenced to in [1] and attached it. The first question is: where shall I put the sources to the library? Somewhere in the checktestdata repository (and if so, where), or maybe create a checktestdata2 repository? Let me give a little more information on why I think an EDSL is a good idea here, and why I choose for Haskell (other than my background): + With an EDSL, you get so much for free from the underlying language, such as (typed) variables, all sorts of abstraction mechanisms, compiler optimizations, etc. For example, some testdata script may consist of some repeated parts, which in an EDSL can simply be given a name to get rid of the duplication. + Compiler optimizations, which of course also exist for C++ in which checktestdata is built now, don't only optimize the checktestdata runtime, but also the script itself since it is just a program and not some custom script anymore. Especially with a functional language such as Haskell this means definitions of primitives can be inlined and whole parts of the script may get fused to a very efficient version. + Maintainability: my library is now +- 200 LOC, including docs and empty lines, and contains quite a lot of the checktestdata language already. Because Haskell has many well optimized parsing libraries, we get a lot for free, and only have to create a nice generalization for this particular use case. + Efficiency: the current checktestdata is just too slow, as mentioned in [1]. As a reference, on my machine the checktestdata script for interception 09.max_ans.in takes 35.5s, the python script takes 2.2s, and the attached version only 2s. And I haven't done anything to optimize specifically for this case, I've just translated the ctd script to the syntax of my EDSL. + With this, the script can do more than just checking the syntax. I.e. when the problem specifies that it contains a connected graph, one can check the syntax, but also the connectedness of the graph, while still separating concerns, i.e. the syntax of the input can be separated from the semantic checks, but the semantic checks don't need to come in a separate program which also has to read the input again, like we should now. All can be combined naturally. + Create your own combinators: the primitives can be easily combined into more complex combinators which you can then reuse. I.e. think of a 'point2d' combinator that reads two space separated integers. - Haskell Platform needs to be available, this is easy on all major OS, but is a quite large dependency. Then again, GCC and Boost are large as well, and are less Windows-friendly. - Slighly different syntax: although Haskell is quite liberal in its syntax for EDSL's, users have to get used to particularities of the language/ - Data generation is not easily possible, if not infeasible. There should be some way of embedding the language to make this possible, and this is actually what I've been working on for the past few years when thinking about this problem, but since we mostly use it for checking data I went with a monadic embedding which is very easy for data checking. - Backwards compatibility: parsing existing scripts and running it with the new backend is not going to be easy. I don't expect show stoppers, but it definitely won't be as efficient as writing scripts within the language because of variable bindings. I need to work on this a bit to be sure though. That's it, of course my libary is not done yet (no good error messages, no regex yet), but it's a proof of concept that is already usable for simpler problems. I hope to convince you that this is the way to go, and to get some ideas to create the POC into a working tool. Also, suggestions on 'interesting use cases' are welcome, the interception is a nice one but I am sure there are many more problems for which the standard checktestdata just not works. Yours truly, Jeroen [1] https://github.com/DOMjudge/checktestdata/issues/3

2 10

Storing participant program in execdir after judging is complete
by Parth Mittal 28 May '17

28 May '17

Hello We run an instance of domjudge for our university courses, which has a lot of submissions (the judging run table has almost a million rows now). I was exploring ways to reduce the disk usage of the judge, and I found that in the directory for every judging run, we store a copy of the executable under execdir under every testcase. Each executable is > 1 MB, and when this multiplies with number of test cases * number of judgings, this becomes very large in our use case. Is there a reason why the executables are stored after judging is complete? Thank you Parth Mittal

2 1

Option to stop judging when met WA
by Phúc Khang Nguyễn 27 May '17

27 May '17

Hi, I'm wondering if there are any options that can stop the judge when it met a Wrong Answer verdict like Time-limit Exceed or not? (maybe change the code perhaps) If yes, please show me how to do it. Best regards, Khang

2 1

Presence at the World Finals
by Jaap Eldering 13 May '17

13 May '17

Hi all, Most of the DOMjudge developers will again be present at the ICPC World Finals, starting in a week in Rapid City, USA. If anyone else is interested to talk and meet, please do get in touch or simply find us at the WFs! We love to talk and get feedback, discuss your use cases, issues, feature (requests), etc. See you in Rapid City! Jaap

1 0

error with judgedaemon
by Fredy Vanegas 18 Apr '17

18 Apr '17

Hi, i am try install the domjudge, domserver and judgehost in the same machine, but when i execute the judgedaemon "./judgedaemon" in the path ~/domjudge/judgehost/bin give me the next error: [Apr 18 13:22:11] judgedaemon[5016]: Judge started on dom-judge [DOMjudge/5.1.3] [Apr 18 13:22:11] judgedaemon[5016]: Not using cgroups. Using cgroups is highly recommended. See the manual for details. [Apr 18 13:22:11] judgedaemon[5016]: Registering judgehost on endpoint http://localhost/domjudge/api [Apr 18 13:22:12] judgedaemon[5016]: error: Error while executing curl POST to url http://localhost/domjudge/api/judgehosts: http status code: 403, response: <!DOCTYPE html> <html lang="en" xml:lang="en"> <head>  <meta charset="utf-8"/> <title>Login failed</title> <link rel="icon" href="../images/favicon.png" type="image/png" /> <link rel="stylesheet" href="../style.css" type="text/css" /> <link rel="stylesheet" href="style_jury.css" type="text/css" /> <script type="text/javascript" src="../js/jquery.min.js"></script> <script type="text/javascript" src="../js/jury.js"></script> <script type="text/javascript" src="../js/sorttable.js"></script> <script type="text/javascript" src="../js/domjudge.js"></script> </head> <body onload="setInterval('updateMenu(0, 0, 0)', 20000); updateMenu(0,0,0)"> <h1>Not Authenticated</h1> <p>Invalid username or password supplied. Please try again or contact a staff member.</p> </body> </html> I appreciate all the help you can give me.

2 1

Dependent problems
by Solal Pirelli 18 Apr '17

18 Apr '17

Hi all, At the Helvetic Coding Contest - as its name implies, Switzerland's largest programming contest - we patched DOMJudge to add support for hiding problem statements until another problem is solved. For instance, if C depends on B and B depends on A, teams are forced to first solve A, then B, then C. This year we had 5 such chains, for a total of 15 problems. This has two advantages: - It forces beginners to solve easy problems first, ensuring they'll solve a few problems and have fun rather than picking a hard problem whose difficulty is not apparent and getting stuck on it for hours. (This looks like a problem that should be solved by educating users, but as any software engineer knows, that is all but impossible :) ) - It allows us to have a bunch of problems on a common theme, where in a few cases the harder one's statement is simply "like the easier one, but with much larger bounds on the input" without allowing teams to solve the hard one first and then submit that solution to the easier one. More problems, less work for problem creators. Thus our contest is suitable for both beginners and experts (even if beginners don't have a chance of actually winning, they enjoy it). I recently rewrote that patch on top of the master branch of the DOMJudge repo, it's quite small (about 40 modified lines). Is this something DOMJudge users are interested in? Should I open a PR? Cheers, Solal Pirelli

3 2