Skip to content

Latest commit

 

History

History
78 lines (61 loc) · 4.55 KB

README.md

File metadata and controls

78 lines (61 loc) · 4.55 KB
Raw

problem-tools

Tools to develop and build competitive programming problems. Should be included as a submodule in problem repositories.

Overview

In repositories using these tools, problems will be grouped by course and by contest. Courses are top-level directories, and also contain templates for the problem statements and solution notes. Contests then are directories in a course directory and contain problem directories.

Each problem consist of:

  • a problem statement written in LaTeX
  • a testcase generator
  • one or more solutions, including intentionally wrong or slow ones
  • (optional) a validator to check submission output for correctness
  • (optional) an interactor for interactive problems
  • (optional) an answer generator if the .ans files do not correspond to the solutions output
  • solution notes created in Ipe
  • a problem.json containing metadata for the problem list
  • a domjudge-problem.ini containing the timelimit

All of the executables are grouped in a executables directory, and can be written in either C++20 or Python 3.

Requirements

To use these tools, you will need:

  • Bash 4+ with the GNU realpath tool installed
  • Make
  • Python 3.7+ with everything listed in requirements.txt installed (run pip install -r requirements.txt with the pip version corresponding to your python 3.7+ version)
  • GCC 8.1 or newer (on macOS, you will need to install this using homebrew, the builtin g++ program links to clang). See the faq entry below for g++ vs. clang.
  • Latex including latexmk (already included with most latex setups)
  • Ipe (the ipetoipe program should be available in your PATH)
  • (optional) GNU parallel to speed up some scripts

For some helper scripts you will also need pdfjam and pdfinfo. Lastly, if you want to run a local judge setup for testing, you will also need the requirements listed in local-judge/README.md.

Usage

You interact with these tools through the setup-problem.py script in the root folder and through the makefiles in both the contest and problem directories. The targets defined by these are described in full detail in docs/makefile.md and docs/contest-makefile.md respectively. You are encouraged to at least skim through these before working on a problem, as many common problems are already solved by these.

The different parts of a problem listed in Overview are described in more detail in docs/structure-of-a-problem.md. You can read through it while building your first problem.

FAQ

  • Why do I need g++ instead of clang? Clang's standard library does not implement ios_base::sync_with_stdio, which means that IO will be significantly slower. This makes it much harder to compare local runtime with runtime on the judge.
  • How do I upgrade old problems to a new format? The setup-problem.py has an upgrade mode (setup-problem.py --upgrade) that aims to help with that process. Currently, it prompts you for problem.json metadata if it is missing, and detects old domjudge-problem.ini's containing outdated/unnecessary data.
  • How do I update the python dependencies? The dependency management in this repository follows the method described here. The packages should be installed in a virtual environment. In summary: the file requirements-to-freeze.txt describes the packages directly required by this repository, possibly with version specifications. The file requirements.txt describes all direct and indirect dependencies, down to the specific version. Thus, running pip install -r requirements.txt ensures identical environments with identical versions. If you want to modify the dependencies, then optionally specify the new direct dependencies in requirements-to-freeze.txt and run pip install -r requirements-to-freeze.txt --upgrade. Now your environment is up to date. To reflect these changes to requirements.txt, run pip freeze > requirements.txt.

Setting up a problem repository

To setup a new problem repository, create an empty git repo and then, from its root directory:

  • Add this repo as submodule in the tools folder: 
    git submodule add [email protected]:compprog-lecture-tools/problem-tools.git tools
  • Create a convenience symlink to the setup-problem.py script: 
    ln -s ./tools/setup-problem.py setup-problem.py
  • Add /timefactor and /login.toml to your repos .gitignore