Here are some updates from the parser & visual editor front from the last few weeks.
The wikitech-l list has had some relevant discussion around:
* a wiki object model: http://lists.wikimedia.org/pipermail/wikitech-l/2011-August/054499.html
* deprecated wikitext markup: http://lists.wikimedia.org/pipermail/wikitech-l/2011-August/054603.html
* the template system: http://lists.wikimedia.org/pipermail/wikitech-l/2011-August/054649.html
The visual editor folks are in a research and development phase, exploring other systems and techniques (research) and prototyping a proof of concept (development).
Trevor and Inez have been making substantial progress on the visual editor. Check out their work: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/author/tparscal and http://www.mediawiki.org/wiki/Special:Code/MediaWiki/author/inez .
They spent part of August working on the transaction protocol/blocking piece. There are two possible models: 1. we save the entire doc at every keystroke, 2. we build it as a series of events, and if we want to undo, we can just reverse it.
Insert, move, & annotate are now reversible. They are replacing the current system with a transaction-based system, which will give us better collaboration, & better undo for free as well. But this is still in research, very much in flux.
Neil is working on a demo of the Etherpad integration with MediaWiki, but much of his time has been taken up with working on the Upload Wizard, as you can see in http://www.mediawiki.org/wiki/Special:Code/MediaWiki/author/neilk . He hopes to get it done ASAP.
We're interested in seeing how MediaWiki's trunk developers can reuse existing Wikia code. Wikia and Wikimedia Foundation have been collaborating and sharing ideas, code, and designs.
Since chat and collaborative editing share auth requirements, Neil wrote an Identity API extension for that. I believe this is it: /branches/extensions-realtime/IdentityApi
From the notes I'm seeing, parser progress is a blocker for visual
editor progress right now -- maybe Brion can speak more to that?
In version one of the visual editor, should we have lots of broken features, or a few working features? It looks like we'll start by picking a set of use cases and supporting only those for a version 1. For example, for new page creation, we don't even need the parser. And Trevor really wants to get something out in the wild to see what people do with it.
Almost all of this is from meeting notes so I may have some things wrong -- please correct me if I do!
We do continue to need help from people with significant experience in user facing, highly interactive applications and with serious JavaScript skills. If you fit that description and can spare some time, take a look at the code links above and reply to the list, or to Trevor, Neil, or Brion.