Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
266 lines (188 loc) · 9.05 KB

CONTRIBUTING.md

File metadata and controls

266 lines (188 loc) · 9.05 KB

Contributing

Table of Contents

Preparing for Contributions

Installing

Prerequisites

⚠ Important (Windows-Users): Configure your editor or IDE to use Unix-style line-endings (LF) for this project. Unix-Style-Endings are required at least for Markdown, JS and JSON files.

git clone https://github.com/<your-name>/glossarify-md
cd glossarify-md
./install-git-hooks.sh  // optional; Run tests prior to pushing.
npm install
cd ./test
npm install

⚠ Important: Make sure to run npm test successfully on a fresh clone before beginning to code. If tests fail (which should not happen but might happen due to unforseen platform issues or bugs) consider filing an issue, first, and hold back contributions until the issue can be resolved. DO NOT just ignore it. DO NOT commit new baselines in this case. Otherwise it will corrupt the baseline and while the CI pipeline might now go green any tests have lost their value.

Experiments

Should you need to try or debug changes to glossarify-md source code then never change the test suite for this. Instead set up an experiment with a configuration like this:

${workspace}/
   |- bin
   |- doc
   |- experiment/          <-- create
   |   |- input/
   |   |   |- document.md
   |   |   |- glossary.md
   |   |- output/
   |   |- glossarify-md.conf.json
   |- ...
   `- package.json

glossarify-md.conf.json

{
    "$schema": "../conf/v5/schema.json",
    "baseDir": "./input",
    "outDir": "../output"
}

Then to run the experiment use...

npm run experiment

☛ Note: If you need to create files outside the experiment/ folder, consider using a *.gitignore.* filename pattern to exclude it from revision control and accidental commits.

Debugging

The project includes a VSCode Debug Experiment launch configuration for debugging experiments with the internal debugger (F5 key).

Alternatively

npm run debug

starts a debug session for an experiment at 127.0.0.1:9229 and allows other debugger frontends to connect to the process. For example you can connect with

  • Chrome Browser ⇨ URL-Bar: chrome://inspect
  • VSCode ⇨ Launch Config Debug External
  • others

To run a configuration at arbitrary location use

npm run dconfig ./path/to/glossarify-md.conf.json  (debug session on)
npm run config  ./path/to/glossarify-md.conf.json  (no debug session)

Testing

Any scripts, paths and examples in this section assume you're working in ${workspace}/test.

cd ./test
npm install  (if not done yet)

Test directories:

${workspace}/
   |- ...
   |- test/
   |   |- input/            <-- test suite
   |   |- output-actual/    <-- actual results from last run
   |   |- output-expected/  <-- currently accepted baseline
   |   |- node_modules      <-- installed test dependencies
   |   `- package.json      <-- test dependencies and scripts
   |- ...
   `- package.json          <-- project dependencies

Baseline Testing

Folder ./output-expected contains the baseline which ./output-actual is compared against (using git diff). If there are any differences between the actual results and the expected baseline then tests fail. Adding additional tests requires the baseline to be updated. It must only be updated if all the the differences are an expected consequence of the latest modifications to the system under test or test suite.

Running the Test Suite

npm test

Note: The script in ${workspace}/package.json will run a linter prior to running the test suite. The script in ${workspace}/test/package.json won't run the linter.

You can run a single test or selected set of tests with npm run at [glob-pattern]. If you only know the test case directory but not the exact run-script npm run at? [directory|pattern] might help you.

Examples:

npm run at? export
npm run at test-p10-13

Extending the Test Suite

A test case usually consists of

  1. one or more document files... ...with term usages in sentences that specify the test case and acceptance criteria in a GIVEN <condition> [AND|OR <condition>] THEN <expectation> [AND|OR <expectation>] syntax
  2. one or more glossary files ... ...with term definitions required for the test goal
  3. a config file... ...whose glossaries section points to the glossary file(s)

To add a new test case or scenario foo-test

  1. create a new feature test directory ./input/foo-test 
    mkdir -p ./input/foo-test
  2. copy and tailor a config from an existing test. In particular adjust outDir and other paths to output-actual!
  3. append a new test_* script in ${workspace}/test/package.json 
    node . --config=./input/foo-test/glossarify-md.conf.json

Important: Each test case must have a dedicated configuration and input file set. Avoid copying existing tests. If you can't resist, though, make sure to adjust any paths in the config.

./input/foo-test/glossarify-md.conf.json (sample)

{
    "$schema": "../../../conf/v5/schema.json",
    "baseDir": ".",
    "outDir": "../../output-actual/foo-test",
    "dev": {
        "termsFile": "../../output-actual/foo-test/terms.json"
    }
}

Review

After modifying the sources and extending the test suite run the tests again. After adding tests the test suite is expected to fail (see baseline testing).

Carefully review the diff of failing tests:

  • the diff should be minimal and only change what you would expect to change
  • changes to other files apart from the newly added test input files are signs of unexpected consequences or a breaking change.

Tweak the implementation & rerun tests as often as necessary.

Once the diff is okay...

  1. Run npm test from ${workspace} root or at least run the linter with

    npm run linter

  2. Commit changes to JavaScript sources using conventional commit messages

    • Bugfixes: fix: <issue>. Closes #<issue-nr>.
    • Features: feat: <description>. Closes #<issue-nr>.
    • Breaking Changes: 
      feat: or fix: ...

BREAKING CHANGE: <description>

  3. Commit changes to the test suite ${workspace}/test/input/* with a message:

    test: New test cases.

Update Baseline

Run the complete test suite to build a fresh ./output-actual

npm test

⚠ Warning: Committing a corrupt baseline can render the test suite useless. Only proceed if you are sure about the contents in ./output-actual and that the Diff is okay [^why-this].

  1. Create a new baseline from ./output-actual with

    npm run new-baseline      (Linux, Mac, Unix)
npm run new-baseline-win  (Windows)

    ☛ Note: At this point you can still restore the old baseline by running

    git checkout ./output-expected

  2. Commit a new baseline

    npm run commit-baseline

    ☛ Important Always commit baselines, separately. This allows for restoring a previous baselines using git reset or git revert.

    ☛ Note: If you have already staged files for commit when running the command but didn't git commit them yet, then these files will be unstaged temporarily using git reset. You won't lose any changes. Just check your git status and git add them again if needed.

  3. git push changes to a remote branch

    ☛ Note: Always use git revert to undo commits if you have already git push-ed to a remote branch.

  4. Open a Pull Request in the origin repository (use dedicated pull requests for different fixes or features)

[^why-this]: This warning is there to raise your care but not to make you feel intimidated. Nothing is broken until a Pull Request makes it into the master branch and if things go wrong git will be our friend.