Automate the retrieval of w3c tests.

[jdalton]

This PR automates the retrieval of w3c tests by an entry in the package.json

  "web-platform-tests": {
    "repo": "w3c/web-platform-tests#ba54984aa8838342ae97aed1fc7bd3cb5499f637",
    "path": "./tests/w3c"
  }

[arschmitz]

👍 This keeps the repo a lot lighter / cleaner I like pulling in the tests

[jdalton]

I'll fix those linting issues.

[arschmitz]

@jdalton look like you need to sign the CLA too

[jdalton]

Ok, signed the CLA.

I'm not sure why the other test fails are occurring, they pass when doing grunt test.

[scottgonzalez]

I'm not a big fan of relying on someone's personal token for this task to work.

[jdalton]

It's a no-scope public token (so harmless, readonly access). It allows getting the higher 50k per hour request limit.

[scottgonzalez]

I still think this change isn't necessary, but I'll defer to the rest of the team.

[jdalton]

@scottgonzalez

I still think this change isn't necessary, but I'll defer to the rest of the team.

What's nice about this approach is:

• It's explicit about what version of the w3c tests PEP is testing against via the sha
• It's easily customizable with configurable repo/sha/path from the package.json
• It's quick with a one time, ~500k download at ~3 seconds, vs. a 60mb+ submodule/dev-dep
• It doesn't require copy/pasting, manual modification, or tracking the w3c test source in the repo

[narqo]

I know that this is opinionated but it looks like you have to download a half of the internet while getting request. Why not to use something more lighter for such a trivial task? How about got or asker?

[jdalton]

@narqo

but it looks like you have to download a half of the internet

That's a bit of a stretch. It dependencies aren't egregious and it only takes ~5 seconds to download.

Why not to use something more

Because I've used request and I trust it, it's the 4th (going on 3rd) most depended on package in npm, and it's a dev dep to boot (so not installed very frequently and only when deving).