On wikitext and wysiwygs
Let me begin by saying I love Wikitext. It’s not pretty, but hell if it doesn’t get the job done. There’s been a lot of talk floating around (especially in light of this grant to improve usability in MediaWiki, which I am 100% for) about trying to get a WYSIWYG working–to acceptable standards–to help lower the bar to contribution.
I hate WYSIWYGs. Granted, I’m also a developer, but I’ve never felt that a WYSIWYG ever gets you as close to what you want as editing the raw source (and hence, I’d be opposed to any solution that took away raw editing and forced a WYSIWYG). I’m not the first to say it, but WYSIWYGs are a misnomer. You will never “get” what you “saw.” However, you’ll get a pretty good approximation of what you meant for it to do. I suppose for basic editing, this is good enough. Bold, italic, that kind of thing is easy to reimplement in a new interface for editing. It gets a whole lot harder when you start adding more complex things, such as tables, templates and other block-level elements.
Unfortunately, without a well-defined grammar (see bug 7), it gets really really hard to do back-and-forth translation of wikitext. Right now, wikitext parsing is a 1 way move. We can turn wikitext into XHTML, but we cannot go the other way. Given a block of arbitrary XHTML, we cannot turn it into appropriate wikitext. This is a bad thing and makes developing a WYSIWYWLTG (What-you-see-is-what-you’d-like-to-get) much harder.
In other wiki news, we’ve got a nice discussion going on on a previous post. Trying to make the ForeignApiRepo more fault-tolerant. Right now, when it fails, it fails hard. Graceful failings are nicer to end users.
What about focusing on a small subset of Wikitext? After all, inexperienced users do not need very advanced features anyway.
For a proof of concept see http://www.mediawiki.org/wiki/Extension:MeanEditor
I think WYSIWYM (intuitive syntax highlighting, like WikEd but without the feature creep) plus live preview will be the way to go. On one hand, it is still somewhat harder to learn than a WYSIWYG, on the other, a WYSIWYG would widen the gap between users who can use advanced wikicodes (template parameters, html etc.) and those who can’t, so WYSIWYM sort of empowers users, which I think is much closer to the the wiki way than the traditional concept of user-friendliness.
Re: Jacopo
Seems promising. Not sure about the patch going into core (although I’m far from being the final authority). It seems rather single-purpose at a quick glance, not useful for much outside MeanEditor.
Re: Tgr
This is true. Making LivePreview not suck would be a great step in the right direction. That plus a WYSIWYM editor dealing with a subset of wikitext (the simple everyday things) could work. Haven’t played with Jacopo’s extension yet, however.
Yes, the patch will need some rewriting before being proposed. A good framework for visual editor extensions will be very useful, though.