Remove JSPM related dep, require it to be installed by the user

[MoOx]

Or more as a peerDep (and require() only if in a jspm context). We need to find a way to import that in a jspm context. Any jspm user can help us about that?
Is pkg-resolve really required?

[koistya]

postcss-import currently doesn't work on Windows, see: babel/babel-loader#251

[MoOx]

PR welcome. We can probably just use the problematic dep as an optional one, should not take that long.

[RyanZim]

@MoOx Any reason not to go ahead with making this an optional dep?

[MoOx]

No, just time I guess. Go ahead!

[RyanZim]

@MoOx Thinking about this more, I'm not sure what this will solve?

If a dependency can be used, but you would like npm to proceed if it cannot be found or fails to install, then you may put it in the optionalDependencies object. This is a map of package name to version or url, just like the dependencies object. The difference is that build failures do not cause installation to fail.

https://docs.npmjs.com/files/package.json#optionaldependencies

pkg-resolve installs OK, just doesn't run OK with webpack in the mix.

I've got a notion to add a jspm option, false by default, that enables requireing pkg-resolve when it is true. This would be a breaking change, but how many people use jspm?

[MoOx]

Probably make sense. I never used jspm so can't really tell :)

[RyanZim]

I never used jspm so can't really tell :)

Me neither!

What do you think of the jspm option? Perhaps we could add it, with the default of true, and change the default when make the next major.

@guybedford Any advice?

[RyanZim]

@MoOx @ai Thoughts on the jspm option? The more I think about it, the more I like it.

[wtgtybhertgeghgtwtg]

It might be a little extreme, but what if you spun off the jspm version into a separate package? It seems like it's been the root of a lot of trouble. Also, it would simplify the dependency graph. Most of the dependencies in postcss-import are from 'pkg-resolve', and not by a small amount.