Plans for cross-browser normalization for contenteditable

[balupton]

How lightweight will Hallo be? In other words:

• Will it bother about the cross browser differences?
• Will it bother with only HTML5 browsers? or HTML4 too?

[bergie]

I think the focus will be on modern browsers. So unless somebody wants to bother with the efforts of legacy IE support, it won't be there.

As for cross-browser differences in generated HTML, the initial plan is to use jQuery clean (http://code.google.com/p/jquery-clean/) in getContents, and probably also for paste events.

[balupton]

https://github.com/balupton/html5edit would be a good low-level interface for this (which could then implement jquery clean if needed), allowing Hallo to focus just on the UI

[bergie]

+1, this is probably the right way to go

[BladeBronson]

I'm very interested in this project. Even the basics seem to be in great need of normalization. For instance, pressing enter twice to create a new "paragraph" will only create a paragraph if already in a paragraph - otherwise a series of divs are used. Should this level of normalization be a goal for Hallo?

[bergie]

@BladeBronson: cross-browser normalization is indeed important. I think that falls under the jurisdiction of @balupton 's html5edit in the scheme where html5edit is the contentEditable core, and Hallo is the UI

[bergie]

Hallo's UX is heavily built on the concept of selections. When you're writing content, you're just writing. When you make a selection, a popover toolbar appears underneath. This popover should only display the commands relevant to the current selection (following rules around inline vs. block-level elements, etc).

Because of this, the lower level library should provide us the APIs and events that inform us of selections, actions possible for them, and allow us to execute these actions.

The actions will include stuff like bold and italic, changing the blocklevel element, and inserting content like images, links, and RDFa-marked annotated content (like persons).

if html5edit can do this for us, that will make life a lot easier for Hallo and practically any other rich text editor out there.

[bergie]

The idea of Hallo is to provide an easily usable rich text editor that has a permissive license, and which is as easy to use as any jQuery Ui widget. A major part of this is a good and universally functioning way to handle the actual HTML formattings, regardless of which browser is used to make them, or whether they come through copy-paste.

Since this is a quite large problem space, it makes sense to join forces here with other editor projects. There are some projects that explicitly seek to aim a rich text editor core that can be used with different shells (which Hallo would then be). Here is a non-exhaustive list of them. Editor cores that have an incompatible licensing model are not considered:

• HTML5 Edit - very interesting handling of selections, but doesn't appear to be actively maintained (ping @balupton)
• wysihtml5 - actively maintained and with a focus on secure mark-up generation, but uses IFRAMEs instead of actual in-line editing
• Substance Surface - very early stage library with a promising approach based on operational transformation, which would make collaborative editing easy to implement
• Squire - another editor core built on IFRAMEs
• Rizzoma - another operational transformation editor core, but with all comments in Russian

[bergie]

Here are some notes on what we would need from Surface: https://github.com/substance/surface/issues/2#issuecomment-6901969

[bergie]

http://yields.github.io/editable/index.html by @yields is another one, based on the web components idea.