Jump to content

Wikipedia:Wikipedia Signpost/2010-09-06/Technology report

From Wikipedia, the free encyclopedia
Technology report

Development transparency, resource loading, GSoC: extension management

Development team to start publishing updates regularly

As the MediaWiki software behind Wikimedia sites grows and matures, it becomes more complicated to manage and to oversee major changes. As such, the Foundation has begun to bring in more paid contractors and employees (though not many for such a large and popular set of websites), each with their own project. The first in a soon-to-be-monthly series of posts outlining these projects was posted this week on the Wikimedia Techblog. The projects that receive some sort of paid support rather than being left entirely to the community to develop include the following. This is not complete list and the items are numbered only for convenience:

  1. Virginia Data Center – To set up a world-class primary data center for Wikimedia Foundation properties.
  2. Media storage – To re-vamp our media storage architecture to accommodate expected increase in media uploads.
  3. Monitoring – To enhance both ops and public monitoring to (a) notice potential outages sooner, (b) increase transparency to the community, and (c) support the progress-tracking required in the five-year plan.
  4. Article assessment – To collaboratively assess article quality and incorporate reader ratings on Wikipedia.
  5. Pending changes – The enwiki trial.
  6. Liquid threads – An extension that brings threaded discussions capabilities to Wikimedia projects and MediaWiki.
  7. Upload wizard – An extension for MediaWiki that provides an easier way to upload files to Wikimedia Commons, the media library associated with Wikipedia.
  8. Add-media wizard – A gadget to facilitate the insertion of media files into wiki pages. Its development is supported by Kaltura.
  9. Resource loader – To improve the load times for JavaScript and CSS components on any wiki page.
  10. Central notice – CentralNotice is a banner system used for global messaging across Wikimedia projects.
  11. Analytics revamp – To incorporate an analytics solution that can grow and answer questions posed by the Wikimedia movement.
  12. Selenium deployment – To build an automated browser-testing environment for MediaWiki.
  13. Fraud prevention – To focus on integrating new fraud prevention schemes in our credit-card donation pipeline.
  14. CiviCRM upgrade – To upgrade our heavily customized CiviCRMv2 install to a mostly stock CiviCRMv3 install.
  15. Process improvement – To increase transparency and generally organize the Foundation’s engineering efforts more efficiently.

Further information on each, including their current status, is available on the original post. Updates on each should be more accessible in future.

ResourceLoader coming "soon"

Developer Trevor Pascal announced on Twitter that his work on a new ResourceLoader to improve loading speeds on Wikimedia sites had progressed and could now be expected "soon". He went into more detail on the Wikitech-l mailing list, explaining the main features to expect:

He gave the example of a page that would previously require 35 requests (totalling 30kB) now taking just one of 9.4kB. Gains for users on older hardware or mobile devices might be improved even more, he said, since they were being served whole scripts they could do nothing with.

Google Summer of Code: Jeroen De Dauw

We continue a series of articles about this year's Google Summer of Code (GSoC) with student Jeroen De Dauw, who describes his project to develop a system for managing the extensions installed on a wiki (read full blog post):

In terms of Wikimedia sites, developments in this field could improve the turnaround time for extension deployment, but the significant gains will be for spreading extensions to and from other MediaWiki-based sites.

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 many weeks.

  • The cause of disappearing images has been identified as a code flaw that allowed the upload to stick in some places on the servers but not in others, where, for extraordinary reasons (such as the recent server transition), some but not all actions were impossible. The upload process will now be consistent in its approach (revision #72021).
  • A two-day public "hackathon" to get together the WMF tech staff with volunteer engineers for "an old-fashioned face-to-face coding sprint" is in the pipeline. This meeting will be formally announced in September and will take place in October on the east coast of the US (Wikitech-l mailing list).
  • Also in the planning stage is an invitational meeting to discuss plans related to "how data is organized, displayed, captured and analyzed on WMF properties" ("Data Summit"). The invite list is to be kept small, but interested parties should apply to danese at wikimedia dot org for an invite (Wikitech-l mailing list).
  • Systems Administrator Domas Mituzas blogged about how the Wikipedia database system compares with those of other websites, and the upgrade from MySQL versions 4.0 to 5.1.
  • The last of the Wikimedia Foundation wikis have been switched over to the Vector skin. At the same time, a new "global opt-out" was enabled, allowing users to easily switch back to Monobook for multiple projects simultaneously.
  • MediaWiki developer Yaron Koren published a proposal for Making Wikipedia into a database (cf. Signpost coverage of related proposals). The article expanded on statements by Koren that had been quoted in an article in Technology Review ("Wikipedia to Add Meaning to Its Pages"), and was written for Hatilda Harevi’it ("The Fourth Tilde"), the Signpost's sister publication in the Hebrew Wikipedia, where it appeared in Hebrew translation last week. Although Koren's consultancy company "WikiWorks" specializes in Semantic MediaWiki, he uses a simple CSV data format in the proposal.