jQuery 3.0 Ideas

gistfile1.md

Just some early draft thoughts

Increased modularity is a long-standing request of jQuery. 3.0 should deliver on that. Many authors will use jQuery 3.0 in ES6 codebases, and often alongside frameworks that have overlap in functionality.

That said, a majority of developers will use a monolithic build. jQuery 3.0 must aggressively remove API surface area and functionality from this core build to protect developers from performance footguns. The jQuery team has reasonably voiced concern that removing modules means it's less likely for people to upgrade. While some users do attempt jQuery version upgrades, many freeze their version and never revisit it. I'd like to get more data on this phenomenon as I think we're optimizing API deprecation policy for a small audience.

Lastly, it's 2015 and the gap between JavaScript developers and jQuery developers has never been bigger. Let's bridge that gap. We should encourage the developer to use modern DOM APIs (that don't suck) and polyfill as neccessary.

Removals

• Remove animation & effects modules from core.
• Remove any event binding shorthands (change() dblclick() etc).
• Remove bind(), delegate(), etc, migrating folks to on(). (#2288)
• Remove support for custom pseudo-selectors entirely
  • Remove positional selectors. (#293)
  • Remove :visible and :hidden.
  • Remove :not, migrating folks to to not()

Defer to standards

• Remove jQuery.ajax() and provide a polyfill for window.fetch.
• Remove jQuery.Deferred and provide a polyfill for window.Promise.
• Remove jQuery.proxy() and recommend Function.bind()

Performance

• Add a public-facing read(),write() transaction model for DOM (a la FastDOM) to protect against layout thrashing and spurious invalidations. Use it internally, fixing the thrashing from jQuery.support checks and others.
• The cleanData() cycle shouldn't have to run on every element removal.
• Event.fix shouldn't be at the top of profiling results. Do the normalization for event.{which target pageX pageY} lazily as a getter.

A few more notes:

jQuery team has some great tickets closed for 3.0 already: https://github.com/jquery/jquery/issues?utf8=%E2%9C%93&q=is%3Aissue+milestone%3A3.0.0+

Today I discovered selector-native.js which IMHO should be the default in 3.0 instead of Sizzle.

I think animations should be removed from the default build of jQuery. Transitions and keyframe anims satisfy many use cases and have great developer ergonomics. For more complex animations, GSAP and Velocity have good authoring stories. At this point, the use cases best served by jQuery.animate not served by CSS or standalone animation libraries are few.

🔊

I can dig it! I could see this also doable though something like a jQ labs.

New jQuery™ and jQuery Classic®

Good stuff. May well have already been discussed elsewhere, but what about browser support? Evergreen only?

Good thoughts, I support you!