ryejs/rye

A modern, lightweight browser library using ES5 natives

rye's Issues

Extending Rye functions

I've been reading Rye's documentation for 2 hours but can't manage to extend Rye functions as I would with JQuery. For example, what would be Rye's equivalent of this JQuery function?

$.fn.blueBorder = function(){
    this.css({border-color:'blue'})
    return this;
};

Append, prepend, after... should accept rye collection

Cure this.each disease

What if instead of wrapping almost every function body in return this.each ... like other frameworks, we used an approach similar to flour, where all methods originally operate on a single element, but are patched at the end to support collections?

https://github.com/ricardobeat/cake-flour/blob/master/flour.coffee#L125

This would allow us to be more efficient in applying methods to one another, for example toggleClass:

exports.toggleClass = function(name, when){
    return this.each(function(element){
        if (when == null) {
            when = !new Rye(element).hasClass(name)
        }
        new Rye(element)[(when ? 'add' : 'remove') + 'Class'](name)
    })
}

It creates lots of new instances just to have the methods available. With this approach it could look like this:

exports.toggleClass = function(name, when){
    if (when == null) when = hasClass.call(this, name)
    (when ? addClass : removeClass).call(this, name)
}

Implement zen-coding like $.create API

bind()ing an event handler makes me lose access to the matched element

Let's say you have

one.on('click .two', function (e, el) {
    // e.target = can be a descendant
    // el == the delegate root
    // this == xxx
}.bind(xxx))

There is no way you can access the actual element (.two). jQuery exposes it as e.currentTarget (I think), along with e.delegateTarget. Should we manipulate the event object too, maybe also normalize e.keyCode/which while we're at it?

Put modules in their own repos

Necessary for #21. Need to figure out how to

• declare dependencies
• build full version
• modules have their own tests, rye does integration tests

Implement custom event emitters for DOM events (namespaces substitute)

Should have support for jsonp requests

removeListener('click') should remove delegated events?

Source map for minified version

Should be easy, just add the necessary parameters for the Uglify2 minify task.

.append() should clone elements?

When called on a collection > 1

EventEmitter Uncaught TypeError: Cannot call method 'apply' of undefined

Getting this when using .once. At some point "something" (a null value) is left in the handlers array where the function once was:

Haven't tracked it down yet.

Event Emitter

1. This implementation dont allow only one handler to same event. https://github.com/Wolfy87/EventEmitter/blob/master/EventEmitter.js#L91
2. The same implementation accepts multiple events.
   https://github.com/Wolfy87/EventEmitter/blob/master/EventEmitter.js#L181

What do you think about both?

Opera does not support instaceof to detect NodeList

document.querySelectorAll('body') instanceof NodeList false
document.querySelectorAll('body').constructor NodeList

.create() fails if the given html doesn't contain any text nodes

Implementation error, we shouldn't be manipulating the live childNodes list.

Create element with html string

util.each should always have the same argument order

Penso que a gente não deva inverter o comportamento do each.

Quando se tá mexendo em coleções que contém tanto arrays quanto objetos esse comportamento é estranho e confuso.

A sugestão é que a gente implemente todos os each como jQuery ou tire qualquer referência a eles da biblioteca apesar de ser super útil ter um atalho para for (var key in obj) if (_hasOwnProperty.call(obj, key))

keyboard mapping/filtering for events

x.on('keydown[enter] .input', function(e){
    //...
})

x.on('keydown[cmd+shift+a]', function(e){
    //...
})

ryejs.com documentation has wrong util.is argument ordering

The web page states:

util.is(obj, type) ⤳ boolean

Yet the latest master and the downloaded v0.0.1 shows:

function is (kind, obj) {
    return kind.indexOf(type(obj)) >= 0
}

Library request in cdnjs.

Hi ~
There's a library request in cdnjs, and we desire to contain rye .
However, I must to make a check for the files that we need to added.

The file i would like to add is:

lib/rye.js

Is that complete for the user? Or other files required?

Thanks for your great work ✨

Piicksarn

.add() should accept a Rye instance

Should class methods preserve whitespace?

User agents may ignore leading and trailing white space in CDATA attribute values (e.g., " myval " may be interpreted as "myval").
Authors should not declare attribute values with leading or trailing white space.
http://www.w3.org/TR/1999/REC-html401-19991224/types.html#type-cdata

The classes that an HTML element has assigned to it consists of all the classes returned when the value of the class attribute is split on spaces. (Duplicates are ignored.)
http://dev.w3.org/html5/spec/single-page.html#classes

The only argument I could find for preserving whitespace comes from a message by Ian Hixie in 2009: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-July/021653.html

He says I try to make the APIs as minimally invasive as possible. This worry seems unfounded in this case. Since whitespace characters are effectively ignored for every possible use of the className property or classList, there is no technical reason to preserve them.

The only case where I think this could be an issue is this:

var old = el.className
$(el).addClass('x').removeClass('x')
if (old === element.className) // fails

But is it reasonable to expect this to work? classList was created exactly to avoid the brittleness of handling strings, and even using the new API this case fails:

var old = el.classList.toString() // == '   x  '
el.classList.remove('x')
el.classList.add('x')
if (old === el.className) // fails since it now == 'x'

And in most other instances (ex: more classes) this would fail too since the order would change (' x y z' => 'y z x').

The only way this could work is if classList methods enforced uniqueness and order, which they don't because of such concerns on being 'intrusive'. I say we should ignore the madness and normalize whitespace, unless someone can come up with a really good use case for whitespace preservation.

Note: order/duplication doesn't matter since specificity is given by the CSS declaration order, not the class attribute - .one.two applies equally to "one two" or "two one", same goes for .one { ... } .two { ... }.

Also of note: Mozilla's provided shim for classList doesn't preserve whitespace: https://developer.mozilla.org/en-US/docs/DOM/element.classList

insertAdjacentHTML failing on Firefox

Remove request and touch modules

These will be moved to their own repos under the RyeJS org. They might still be maintained, but most people will want to use other more feature-complete alternatives.

Add .invoke() method to collections

Since we don't have a few helper methods like focus(), would be nice to have invoke operate on a collection:

form.find('input').invoke('focus')

It could automatically map() over the elements for return values, unlike attr and other methods that return only the first value:

list.find('li').invoke('getAttribute', 'id') // => ['a', 'b', 'c']

Support AMD / CommonJS / bowser / vala / components

Implement XHR methods

Setup CI using Sauce Labs

API names

Based on this post http://blog.millermedeiros.com/jquery-wish-list I want to suggest some modifications:

• not to exclude
• has to filterHas, ?

Ajax events

The ajax API can have a emitter and emit events trought that. I will try implement this.

Do you have any other ideia?

Implement .remove()