fxa/docs/adr/0009-testing-stacks.md

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

68 строки
2.9 KiB
Markdown
Исходник Обычный вид История

# Consistency in testing tools
- Deciders: Les Orchard, attendees of testing discussion at Berlin All-Hands 2020
- Date: 2020-02-04
## Context and Problem Statement
We have a variety in tools used for unit & functional tests. This variety means developers need to learn & know many different tools. It would be nice to reduce this mental workload, while still using appropriate tools for testing.
## Decision Drivers
- Ensuring we're using the right tool for the job.
- Reducing developer mental workload for working with tests.
## Considered Options
- Convert all tests to use a single consistent stack of testing tools
- Accept ad-hoc testing stacks to be chosen as needed
- Identify a consistent stack of testing tools per aspect of the project
## Decision Outcome
Identify a consistent stack of testing tools per aspect of the project
## Pros and Cons of the Options
### Convert all tests to use a single consistent stack of testing tools
Many of our tests use some arrangement of [Mocha][], [Chai][], and [Sinon][] as a stack. We should just use that everywhere without variation.
- Pros
- Developers only need to learn one stack to work with tests throughout the project
- Cons
- Since we already have a variety of testing stacks, we'd need to rewrite a significant number of tests to assert consistency
### Accept ad-hoc testing stacks to be chosen as needed
Different services & apps in FxA could benefit from different ways of testing. Let's just pick whatever works on an ad-hoc basis.
- Pros
- Developers can choose whatever testing stack seems to make sense for the current task
- Cons
- Further variation of testing stacks over time, requiring more effort to learn how to work with tests in any given spot of the project.
### Identify a consistent stack of testing tools per aspect of the project
For React apps, we use [Jest][] and [React Testing Library][]. This is the stack that comes out of the box with [Create React App][], which makes it kind of a de facto standard for React.
All of our tests for microservices use some arrangement of [Mocha][], [Chai][], and [Sinon][] as a stack. We should keep doing that for future services.
Functional tests run against fxa-content-server use [Intern][] & [Selenium][]. These tests are mature and wouldn't really benefit from rewriting.
Future functional tests _might_ benefit from being written with a newer stack using Selenium or some other browser-automation tool (to be determined).
- Pros
- Although there's more than one set of testing tools, we can minimize the variation.
- Cons
- Developers still need to be aware of more than one set of testing tools.
[jest]: https://jestjs.io/
[react testing library]: https://testing-library.com/docs/react-testing-library/intro
[create react app]: https://github.com/facebook/create-react-app
[mocha]: https://mochajs.org/
[chai]: https://www.chaijs.com/
[sinon]: https://sinonjs.org/
[intern]: https://theintern.io/
[selenium]: https://selenium.dev/