Proposal: automatically treat URLs as empty:

[jwhitley]

In requirejs-rails 0.8.0, I added/documented support for folks doing an r.js build via the gem and using CDNs. One of the cases to be covered was the use of a paths config with URLs in the right-hand side. It occurs to me that I ended up effectively making a policy change to the way r.js works: instead of erroring out on URLs and requiring the build config to explicitly specify empty:, I automatically replace any URLS in paths with empty:. This was mostly in response to the stream of folks who get bit by that on the mailing list and the fact that r.js has no other non-error behavior in this case.

I'm proposing to push this logic up into r.js, turning this issue into a pull request which makes said changes. There are a few ways this could go:

1. Just change the default, optionally emitting a notice whenever URLs are substituted (one notice per run) to reduce surprise by users.
2. Leave the current default but add a build config option that globally allows the more friendly behavior. End users thus have less complicated logic when assembling a build config: use the require.js paths config plus one additional "yes, I mean it, I'm really using a CDN" option.

[jrburke]

Seems reasonable, putting it in the 2.0 bucket as that is the active dev branch at the moment.

[jrburke]

There is a dev2.0 snapshot update with a fix for this. It prints out a message that it is is skipping an url, and should only do it once per build layer. Feel free to give it a spin and confirm the behavior.

[jwhitley]

Awesome. I'm overdue to update requirejs-rails' corresponding dev2.0 branch, so I'll take a look at it over the next few days when I do the update.