And it's clearly not an opinion that many large, modern JavaScript codebases are following this practice effectively. I linked to 'React' itself as proof of this...
Also, I personally never used the term "best practice" however I will say this: (1) these practices are common within some of the most popular JavaScript projects, and (2) having used them and also previously stored tests separately, I do prefer storing tests alongside the files I am testing for the reasons I stated. I am not sure why, but your response felt a bit like you were angry with me, and putting words in my mouth.
> 2) the require/import paths are not like '../../../../../' (which is awkward to reason about)
> or requiring alias resolver configurations (which are environment-specific and fragile)
Compare writing an import path within a test of `import Button from '../Button'` to `import Button from '../../../../src/features/abc/components/Button'`. The latter requires you to count the directories in the path and replace each with '..'. This is a pointless waste of mental energy, and the former solution removes the need to do it.
The other solution as I described is to use things like aliases but this causes you to create complex configuration for your tests [0] which must be maintained and means your test are now dependent on this environment.
If this is just an opinion, please demonstrate that it's incorrect.
> (3) you are not maintaining an extra directory
> structure which could diverge by accident.
You are suggesting to me that `src/features/abc/components/Button.js` and `test/features/abc/components/Button.js` are not separate file-system structures which can diverge?
Because they are separate directory structures it is possible for them to diverge. This is a basic implication from the fact that they are separate structures.
If somebody moves `src/features/abc/components/Button.js` to `src/features/def/components/Button.js` you are telling me that you will not need to make any changes at all to `test/features/abc/components/Button.js`?
