Why is window exposed to required modules?

[natevw]

I'm trying to use a module that tests for a CommonJS instead of browser context via (typeof window === 'undefined').

According to https://github.com/rogerwang/node-webkit/wiki/Differences-of-JavaScript-contexts#wiki-determining-thecontext-ofascript this should work fine, but for some reason this was landed a while ago 7fb1314 and now contradicts the documentation.

How should I be testing for CommonJS? I went with the 'window' check because I figured in the browser who knows what globals (exports,module,global) might have gotten randomly set by preceding scripts, whereas it was (other than this!) less likely that a CommonJS module would have random extra globals hanging around.

Want to back this issue? Post a bounty on it! We accept bounties via Bountysource.

[katanacrimson]

Interesting. What's the properties, etc. of window? Is it -the- webkit window?

[natevw]

Sure seems to be. The commit message says simply "Expose window object to Node's context"

[katanacrimson]

You know, I wonder if there's a better way around this.

@rogerwang - why not set a global - say global.context, with either node or webkit (or blink, whatever) to help identify the current exec context?

[natevw]

A one-off special check is not going to be great in my library, nor I suspect in others, especially since I know of no existing standard here and I'd hate to have to add checks every random different environment like this that comes along. Why not make users opt-in with require('window') or something so that it matches the documentation again?

[jduncanator]

@natevm That certainly sounds like a good idea, however is it feasible? Also, what happens if you have an actual module called "window" you wish to require? What happens then?

[tomasdev]

@rogerwang so... continuing #superagent/95 removing 'window' will be a backwards compatibility challenger.

I don't think the proper check frameworks should do is "is it browser or not" but rather a feature detection to know whether UMD/AMD are present as @jashkenas did at jashkenas/backbone@ab5d2eb and @jdalton did at lodash/lodash@e76685f

In node-webkit, as long as we can still assign variables/functions to the front-end (maybe require('window') as @natevm suggested,) it would be plausible to remove window as a global.

[tomasdev]

@damianb a priori is a placeholder for the webkit's window, a posteriori is it. Basically Node context is initialized before the browser context, AFAIK.

[nwjs-bot]

This should be working with latest version now.

In 0.13 we changed to an optimized architecture so more features can be supported, see http://nwjs.io/blog/whats-new-in-0.13/ and it's good for keeping up with Chromium upstream -- we released with Node.js v6.0 and new Chromium versions within 1 day after upstream release.

The new version would fixed many issues reported here and we're scrubbing them. This issue is closed as we believe it should be fixed. Please leave a message if it isn't and we'll reopen it.