Task Development Tools for the IIOT

This repository contains a selection of task development tools, to be used for the preparation of the rounds of the IIOT.

Setup

All tools should be used in a unix bash: either Linux, MacOS, or WSL under Windows. Task development requires task maker: a setup guide for it, including documentation of the main features, is available online here.

Intended workflow

The preparation of an edition of the IIOT should follow the following workflow.

1. Create a new repository for the edition, through tool new_year.sh (requires gh installed).
2. In this repository, create contests skeletons as they occur through tool new_contest.py
3. In a contest folder, create task skeletons through tool new_task.sh
4. Edit the skeleton task until it matches the intended task. The check/ folder may be erased, if plain white-diff comparison is sufficient for scoring; the other folders are mandatory. You can find a description of the task format here.
5. For preparing the task, you can use make-templates to automatically produce templates for your tasks (see more details below)
6. As you prepare the task, you can test the solutions (including partial and wrong ones) through task maker, running command task-maker-rust
7. Once the task is ready, and all solutions are correctly scored, you can run task-maker-tools add-solution-checks -i to store the result.
8. Every time a push is being made to the main branch, GitHub should update a CMS instance running all tasks in the edition (work in progress).

In order to automatically produce templates, you can install a dedicated tool through:

pip install make-templates

You need to edit the file inout.slide in your task folder, that the new_task.sh script created, to match the input-output format that you want for your problem (you can find here a description of the SLIDe language if needed, but it is mostly self-explanatory). Then you just type:

make-templates

and the tool will prepare all the needed templates, and also the input/output description section in the task statement.

Task format description

Main folder

The main folder only contains the single file task.yaml.orig, with general setup information about the task. Some entries are meant to be updated:

• title, with a meaningful long title
• syllabuslevel, with the actual syllabus level from 1 to 5
• memory_limit, with the wanted memory limit
• time_limit, with the wanted time limit

The other entries are unlikely to be modified.

att folder

This folder contains files that the contestants can download while browsing the task on CMS. By default those include templates in various languages, and sample input/outputs.

check folder

This folder contains a single source file of a checker, that attributes a score (from 0 to 1) to a test output, knowing the test input and the correct test output. Many tasks do not need a checker, as plain white-diff scoring is sufficient: in that case, the folder can be fully omitted. For the remaining tasks, the skeleton includes a general-purpose checker with safe read operations that can be modified to fit various needs.

gen folder

This folder contains the input generation data, split into few files:

• limiti.py with a description of the limits of the various subtasks;
• validator.py with a script that double-checks whether an input is valid;
• generator.py which generates inputs given some command-line parameters;
• GEN which lists the set of parameters to be fed to the generator for each subtask, together with some auxiliary information (subtask score, name and dependencies).

sol folder

This folder contain the full, partial, and wrong solutions that should be checked for correctness. File solution.cpp is used to generate the official output. A Python solution should also be included, in order to know how many points can be scored in Python. Templates are symlinked from the att folder, and should produce wrong answers (without crashing). Further solutions, for subtasks or other reasons, can be added here.

statement folder

This folder contains the task statement. The statement is compiled from a (patched) LaTex source english.tex, which is provided in the skeleton. The statement is compiled to english.pdf by default by task maker, while it tests solutions. You can also build only the statement through command:

task-maker-tools booklet

Pictures to be included in the statement should be added here. Beware that in order to enable booklet compilation, pictures should have unique names across different tasks in a same contest. It is therefore suggested to name pictures taking inspiration from the task name.

Tool list

new_contest.py

Creates a contest main folder. Requires to specify the round number, the year, the start and duration of the contest.

new_task.sh

Creates a task skeleton, given its name.

new_year.sh

Creates an edition repository, given the year.

update_tools.sh

If run in an edition repository, it updates the tools folder to the latest version.

rename_task.sh

If run in an task repository, given its current name and a new one, it renames the task.