Creating the DOM tree with createDocumentFragment() and later add it to the actual DOM will bypass expensive Layout operations for each DOM method (Layouting is done only once during inserting the fragment into the actual DOM). This will speed up things a lot.

Plus, sometimes I want to prepare a DOM fragment and only add it later to the DOM tree. This is only possible with the aforementioned method.

http://jed.github.io/domo/ will redirect to http://domo-js.com/ which returns 403...

STYLE.on('#MY_DIV',{zIndex:10})

Produces:

MY_DIV{z-index:10px;}

Errors in IE 8, appendChild not supported for IE element DispHTMLStyleElement.
Occurs when building style elements.

Please add AMD support.

i'd like to have HTML generation and DOM generation in the same library, to enable use on both server and client, and to allow client rendering to take place using innerHTML where faster.

but.

i am unsure how to implement it. seems to me there are two choices:

1. have two parallel DOM builders DSLs, one that generates HTML and one that generates a DOM, like this:

domo.dom.<nodeName>
domo.html.<nodeName>

1. have one DOM builder DSL that creates lightweight mock DOM objects, with the following methods:

Node.prototype.toDOM = function(document) {}
Node.prototype.toHTML = function() {}

the first is probably more efficient because there is no intermediate object generation, and easier to use because there is less API surface area and DOM objects are returned directly.

the second is probably cleaner, since no code needs to be duplicated for each approach, and creation of intermediate objects would allow more methods to be introduced if/when needed.

This doesn't work, the style text-align "right" doesn't take:

        P({style:'text-align:right'},
            "maybe maybe maybe this works?")