To be honest, given that most people load jQuery from a CDN anyway, all this does is make you trust the library you're using less – the beautiful thing about jQuery is that it works, and you never have to worry about what methods are made available to you.
Really like the idea of this but it is SO far away from '90% of jquery'.
I tried using it on my site (with all parts) and litterally NONE of the jquery worked. There is no hover support, no 'attr', no animation and val doesn't work with checkboxes.
You would have to build a site from scratch with this in mind, it is not a drop-in solution for most people.
I wonder if the savings in script size and parse time matter if jQuery is referenced from a CDN [1]. It may save 500ms (or less depending on the mobile browser) with the small size and parse time, but it may also add 500ms (or more on slow mobile networks) in latency.
It would be great if I could scan my current js and build the jquip based on that. Actually, there should be a tool that removes all unnecessary code that's never called. Might be hard or impossible to get that working reliably though.
Using something like uglify's parse-js.js [1] that gives you back the AST might be the right way to go about it. Just count the calls to each $ method in your code.
