Skip to content

Key terms

Vocabulary for contests, problems, submissions and judges as used in the docs, API and web UI.

TermMeaning
ProblemPrompt + resource limits + test cases. Has numeric id and string identifier. Public problems appear in the practice catalog; private ones are contest-only or admin-only.
Test caseInput/output pair. Hidden cases exist for formal grading. Judge stops at first failure (ICPC-style).
SubmissionCode + language for a problem. Lifecycle: PENDING → verdict. May include event_id when submitted during a contest.
EventAPI name for a contest: timed problems + registration + standings. UI says contest.
Event problemA problem attached to a contest. Has its own numeric id within the event (distinct from global problem id).
Custom input runExecute against stdin without grading. The editor Run button. API: input_submissions.
UserAccount. Admin manages problems/contests. Everyone else submits.
LanguageRuntime in judge image + DB row. Matched by name between languages.toml and Postgres.
ClarificationQuestion asked during a contest; answered questions become public clarifications.
NotificationIn-app alert for new questions or answers on contests you participate in.

Problems and contest attachments store three limit fields:

FieldMeaning
accept_timeoutWall-clock budget for judging (stored on problem/event)
execution_timeoutCPU/run budget (stored on problem/event)
memory_limitMemory cap in MB (stored on problem/event)

These appear in admin forms and the API. The judge currently applies fixed nsjail defaults (10s run CPU, ~6 MB virtual memory) — limits on problems are not yet passed into the sandbox. See Judge service — resource limits.

StatusMeaning
PENDINGIn queue or running. Normal for a few seconds.
ACCEPTEDAll tests passed.
WRONG_ANSWEROutput mismatch.
TIME_LIMIT_EXCEEDEDToo slow.
MEMORY_LIMIT_EXCEEDEDToo much RAM.
RUNTIME_ERRORCrash / nonzero exit.
COMPILE_TIME_ERRORDidn’t build. Check stderr.

Standings rank participants by:

  1. Problems solved (more is better)
  2. Total penalty time (lower is better) — solve minutes + 20 minutes per wrong attempt before AC
  3. Total submission count (fewer is better)

Team contests require team registration to submit; rankings are still per user. Full guide: Run a contest.

How judging works: Judge service.