Update browser support check

[koenpunt]

Old

  browser_is_supported: function() {
    if (/iP(od|hone)/i.test(window.navigator.userAgent)) {
      return false;
    }
    if (/Android/i.test(window.navigator.userAgent)) {
      if (/Mobile/i.test(window.navigator.userAgent)) {
        return false;
      }
    }
    if (/IEMobile/i.test(window.navigator.userAgent)) {
      return false;
    }
    if (/Windows Phone/i.test(window.navigator.userAgent)) {
      return false;
    }
    if (/BlackBerry/i.test(window.navigator.userAgent)) {
      return false;
    }
    if (/BB10/i.test(window.navigator.userAgent)) {
      return false;
    }
    if (window.navigator.appName === "Microsoft Internet Explorer") {
      return document.documentMode >= 8;
    }
    return true;
  }

New

  browser_is_supported: function() {
    if ("Microsoft Internet Explorer" === window.navigator.appName) {
      return document.documentMode >= 8;
    }
    if (/iP(od|hone)/i.test(window.navigator.userAgent) || /IEMobile/i.test(window.navigator.userAgent) || /Windows Phone/i.test(window.navigator.userAgent) || /BlackBerry/i.test(window.navigator.userAgent) || /BB10/i.test(window.navigator.userAgent) || /Android.*Mobile/i.test(window.navigator.userAgent)) {
      return false;
    }
    return true;
  }

[Pezmc]

[nb] Why not /Android/i.test(window.navigator.userAgent) and /Mobile/i.test(window.navigator.userAgent); here?
I find it slightly harder to skim read what's going on like this.

[koenpunt]

(some sort of) consistency; this way all tests are on their own line

[tjschuck]

even though if/else tends to be faster, the resulting js of this switch is more concise

Should we care about the resulting JS?

[koenpunt]

I'm actually not sure if if/else in JS is faster than a switch, but I know that it's the case in some languages.

[koenpunt]

And I assumed the resulting JS to be less bytes, but it turns out that's not the case either 🤔

Because UglifyJS apparently can optimize that if to a long chain of ternary operators..

[koenpunt]

I've reverted to an if, with less duplication

[koenpunt]

Maybe we should put the "Microsoft Internet Explorer" test on top, because regexes are more expensive. Although this could also be seen as micro-optimization..

[stof]

@koenpunt moving it to the top is a good idea. It may even help UglifyJS optimizations, as it will see a ìf (...) {return false } return true and be able to shorten it to return !(...)

[koenpunt]

46 characters less than the current:

[tjschuck]

@koenpunt There's a subtle change here — this will no longer match the opposite order, e.g. Mobile blah blah Android, which the original did.

Not sure if anything has a user agent like that, but just pointing out that this is not 100% equivalent.

[koenpunt]

I know, but Mobile always appears after Android according to https://developer.chrome.com/multidevice/user-agent

[tjschuck]

👍

[pfiller]

I tend to prefer readability over micro optimizations that save us a few characters. In this case, however, it seems we get both! I find the changes slightly more readable so I'm 👍

Gonna merge it in. Thanks @koenpunt!