Determines whether PUBLIC_URL is same-origin before registering SW.

[jeffposnick]

R: @Timer @gaearon
FYI: @addyosmani

This is a bit more explicit than trying to use a regular expression. It uses the URL object inside of a block that first checks to see if navigator.serviceWorker is defined; URL should always exist in browsers that support service workers.

From testing locally, I saw scenarios in which process.env.PUBLIC_URL wasn't set at all (which contradicts what @Timer suggested in the issue).

The code in this PR should behave properly regardless of whether it's unset, whether it's set to a local path, or whether it's set to a full URL.

Fixes #2374

[facebook-github-bot]

Thank you for your pull request and welcome to our community. We require contributors to sign our Contributor License Agreement, and we don't seem to have you on file. In order for us to review and merge your code, please sign up at https://code.facebook.com/cla - and if you have received this in error or have any questions, please drop us a line at cla@fb.com. Thanks!

If you are contributing on behalf of someone else (eg your employer): the individual CLA is not sufficient - use https://developers.facebook.com/opensource/cla?type=company instead. Contact cla@fb.com if you have any questions.

[Timer]

From testing locally, I saw scenarios in which process.env.PUBLIC_URL wasn't set at all (which contradicts what @Timer suggested in the issue).

It should always be provided; it just may be blank. This is changing in the near future.

Either way; this code should work.

[Timer]

I'm guessing we will need to add url-polyfill for this?

Would it be better to use Node's url?

url.parse(url.resolve(window.location, process.env.PUBLIC_URL))

> url.parse(url.resolve('https://github.com/facebookincubator/create-react-app/pull/2432', 'http://google.com'))
Url {
  protocol: 'http:',
  slashes: true,
  auth: null,
  host: 'google.com',
  port: null,
  hostname: 'google.com',
  hash: null,
  search: null,
  query: null,
  pathname: '/',
  path: '/',
  href: 'http://google.com/' }

// --

> url.parse(url.resolve('https://github.com/facebookincubator/create-react-app/pull/2432', '/foo'))
Url {
  protocol: 'https:',
  slashes: true,
  auth: null,
  host: 'github.com',
  port: null,
  hostname: 'github.com',
  hash: null,
  search: null,
  query: null,
  pathname: '/foo',
  path: '/foo',
  href: 'https://github.com/foo' }

[jeffposnick]

All browsers that support service workers support the URL object, and the code that uses the URL object is conditionally executed inside of a block that first checks for SW support. So it seemed safe to use the native URL here. I'm happy to throw in a 'URL' in window check if you'd feel better about that.

If we shouldn't use URL for some reason, I'd rather re-implement this using string comparisons than incur the cost of bringing in a polyfill or external library.

[Timer]

All browsers that support service workers support the URL object, and the code that uses the URL object is conditionally executed inside of a block that first checks for SW support.

Oops, I overlooked this. Great!

[gaearon]

Thanks!