User:Brion VIBBER/Berlin roundup
I visited Berlin to drop in on the 22C3 conference and get some face-to-face chatting done with some of our European folks.
Jens has been experimenting with a metadata system, where various plugins can handle data for geographic info, personal infoboxes, etc with their own storage backends and such.
While WikiData might end up being able to handle this stuff in a more fine-grained (if scarier!) fashion, there's some cool infrastructure in this which can be useful either way, such as a message-queueing system for processing backgrounded or offsite data updates.
Magnus has rewritten the page-review code into an extension, and given it an interface overhaul.
Tim's apparently gotten slightly sidetracked on the stable-version stuff with fixing up how we store template links; he's now going to go over Magnus's experimental stable version extension to compare how their versions are working, and we'll see what we wind up with. :)
Gerard's been very active in pursuing content agreements and support for the upcoming Ultimate Wiktionary (or whatever it's going to be called this week). He's got a few organizations on the hook for folding industry-standard-type glossaries and thesauri into the system, with some interested in providing support back to us (servers, development help with integration, etc).
If some of these things go ahead, it'll be important to integrate foreign user authentication for some things. He's recommending looking into the A-Select Authentication System , which is widely used in European educational, government, and financial institutions, and is apparently open source and all that. Allegedly it should be possible to bridge it to OpenID as well, which has been talked about occasionally.
As the UW evangelist, Gerard is pushing it as a way we can solve various problems within the other WM projects as well:
- Interlanguage linking
- Meaning-to-meaning translation info in UW gives the kind of info we'd need for making interlanguage links more cleanly centralized and automatable
- Localizable and semantic categorization and searching for Wikimedia Commons
- Currently category tags are hardcoded strings in some language, not unlike tags on flickr. These are problematic when you have content and users who speak different languages; tying the category tags into a multilingual thesaurus database should make it possible to search for "Pferd" and get a picture of a "Horse", etc.
- MediaWiki localization
- Localized UI messages are basically like anything else that could be in a special-purpose multilingual glossary; a clean, easily-manipulable, (single-sign-on-enhanced!) interface for that could make it a lot easier to deal with keeping the interface localizations complete and consistent across sites.
Gerard gave me a Kennisnet pen and a Kennisnet mobile phone holder, so I'm bought and paid for! ;)
Erik has been hard at work making a Wikidata system for UW actually feasible. At this point he's got a demo set up with GEMET thesaurus data, which is pretty nifty. The data versioning system he's got in the schema currently looks rather weird, but seems like it ought to work; things seem to be coming along nicely.
Another thing he'd like to push is multilingual content at the page level in MediaWiki. This is something we kind of discussed years ago and abandoned, in favor of making a separate database for each language. But for some sites that are oriented as language-neutral or multilingual (meta, commons, wikisource, etc) it could be quite useful to be able to have content in different languages: more integrated than separate databases, but distinct enough that you could filter Recent Changes, mark the HTML lang attribute properly, set RTL mode for Hebrew/Arabic scripts, have pages with the same title spelling, etc. There are some questions about how you'd deal with this exactly, but it's something we probably want to think strongly about.
In the short term, I'm going to merge the namespace manager onto the main branch in the next week or so; this'll let us more cleanly work with additional and aliased namespaces for the sites that keep asking for Portals ;) and will serve as the base for wikidata-like plugs.
Putting up the Review extension to try it live should happen in a similar timeframe.
A 1.6 release should come shortly after this; target early-mid January.
I've agreed with Gerard and Erik to prioritize implementing the single login transfer within the next month or two, and then we can see about external authentication systems (aselect & openid).
Before the final explosion, users without e-mail addresses set will get a warning that it may affect their transfer to single sign-on if their accounts can't be confirmed as matching. This will use a dismissable site message system which I should be getting done early January.
Target for single sign-on: mid-February to early March.