Wikipedia:Wikipedia Signpost/2013-03-25/Technology report

James Forrester updates the Signpost on progress with the Visual Editor
Since its inception in May 2011, the Foundation's Visual Editor (VE) project has grown to become one of its main focuses. Considering that a Wikimedia-deployable WYSIWYG wikitext editor has been one of the most requested features since the language was first used, the hype is perhaps unsurprising. As the project nears its two-year birthday, the Signpost caught up with Visual Editor project manager James Forrester to discuss the progress on the project.


 * Hi James. So to start with the obvious, 18 months in, where are we now?
 * Visual Editor, March 2013.png. The Visual Editor is currently deployed to the English Wikipedia as an opt-in test for all users to be able to edit all articles and user pages; it's also on MediaWiki.org as always-on for the VisualEditor: test namespace. Right now, it supports text, headings, preformatted text, basic annotations (bold, italics) and links (internal and external); other items, such as images and templates, are "alienated" - marked as not-able-to-be-edited and shaded out in green and white. As we add the ability to edit new components ("node handlers"), these green items will gradually disappear.
 * The Visual Editor currently works equally well in both Monobook and Vector and we'll look to keep it that way (though all our design cues are off Vector).


 * July has recently been given as a possible rollout date. What needs to happen before that, and what will be included?
 * We're aiming for the Visual Editor to become the "default" editor - that is, when you click the "edit" button, you get the Visual Editor. There'll still be the current text-based editor, of course, accessed via "Edit Source". In addition to the functionality already present, the aim is currently for us to also deliver all four of templates, references, categories and images (at least, basic abilities of each) before we roll-out.
 * One of the key items before July is to allow templates to be edited in a way that users aren't expected to memorise how they work. To help with that, there will be a new extension ("TemplateData") which will let users add hinting to templates on how they should be used. We'll be deploying that soon and making some initial example ones so that it's clear to community template writers how it will help.
 * In addition, I want to get us testing on non-English Wikipedias so that we know how much more we have to do on i18n/l10n support - we think we're reasonably far, but our users can tell us much more accurately than our pontificating from the ivory tower.


 * So in July it’ll be English Wikipedias and non-English as well?
 * That's the intent. Almost certainly not non-Wikipedias, sadly, as we've not yet had time to look at their specific needs (like integrating Wikisource's ProofreadPage extension).


 * Sounds great. And after July it'll be a case of incrementally adding whatever's still missing -- tables, for example? Are there some things that will never be implemented on the VE, do you think? Magic words, manual interwikis?
 * Yes, table structure editing and other things are on the backlog for 2013/14. It's quite possible that we'll never implement some things, but there are lots of things for us to work on before then. For example, yes, interwikis (individual and Wikidata) and page-setting magic words will both be supported and editable (but probably not for July).


 * Yes. So finally, the project will be two years old in May. If the WMF did it again, what would it do differently?
 * Well, actually we were talking about a rich-text editor in 2002. But yes, obviously we'd love to have made more progress, faster. I think we're confident that we are balancing the desire to get it out as fast as possible to the need to build the best possible editor that we can, so that editors have it easier (and especially for new editors).
 * Ultimately, both the Visual Editor and Parsoid are both research projects still - a good chunk of what we're doing has never been done before, or at best only been done a few times, and there's no settled way to do them. That's why it takes time, and the dead-ends we've encountered and the times we've chosen to change course have led to VE and Parsoid being higher quality and ultimately better for our users


 * Thank you.

In brief
Not all fixes may have gone live to WMF sites at the time of writing; some may not be scheduled to go live for several weeks.
 * MediaWiki 1.21 in the pipeline: Discussions have started regarding the branching and release of MediaWiki 1.21 to external sites (wikitech-l mailing list). Wikimedia wikis have already benefited from the same code, deployed in 12 batches (wmf1 through wmf12) over the last six months. The issue was complicated by the need to briefly rollback wmf12 from Wikimedia wikis this week after the deployment broke page move functionality on setups with multiple databases (bug #46397; also wikitech-l).
 * Wikidata phase 2 coming to a wiki near you: Phase 2 of Wikidata – allowing data points to be replicated from Wikidata.org to client wikis – will go live to ten wikis including the Italian and Russian Wikipedias on 27 March (wikitech-l, blog post) following a successful deployment to a testwiki on Monday. The wikis included collectively account for some 10% of all Wikimedia pageviews.
 * Pick of the blogs: The Wikimedia blog this week carried a round-up of the Foundation's improvements to the translation process for interface messages and other documentation, while former Localisation Team member and serial commentator Gerard Meijssen blogged about the interesting case of trying to create a Wikipedia written almost entirely in sign language. Elsewhere, WMF-mentored Outreach program for Women (OPW) intern Mariya Miteva wrapped up her experience over the last three months. "If you need anything from anyone, ask as early as possible, let them know you will need their help and when, and poke them if you are not hearing back", Miteva advised future interns.