Open
Description
A cloned assignment with autograding has the following .github structure:
.github/classroom/autograding.json <- this file enumerates the tests that will be run
.github/workflows/classroom.yml <- this file runs the tests
So e.g. for me, the classroom.yml is:
name: GitHub Classroom Workflow
on: [push]
jobs:
build:
name: Autograding
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
- uses: actions/setup-python@v2
with:
python-version: 3.8
- name: Install numpy
run: |
pip install numpy
- uses: education/autograding@v1
The magic - uses: education/autograding@v1 line the action to run that
The autograding.json is fully undocumented, but it looks like:
{
"tests": [
{"name": "test 1", ...},
{"name": "test 2", ...},
]
}
each test entry is like:
{
"name": "Compile code",
"setup": "",
"run": "gcc -std=c99 -fopenmp -Wall -Wextra -o simulation simulation.c -lm",
"input": "",
"output": "",
"comparison": "exact",
"timeout": 2,
"points": null
}
and then you have to read https://github.com/education/autograding/blob/master/src/runner.ts to figure out how these things are interpreted
runner.ts
import {spawn, ChildProcess} from 'child_process'
import kill from 'tree-kill'
import {v4 as uuidv4} from 'uuid'
import * as core from '@actions/core'
import {setCheckRunOutput} from './output'
TBH: I think the better thing is to just write a normal workflow run, ignoring github's autograder, that does some scoring assignment
Metadata
Metadata
Assignees
Labels
No labels