The Foundation's Engineering Report for March was published last week on the Wikimedia Techblog, giving a brief overview of all Foundation-sponsored technical operations in the last month. Most of the major developments (Brion Vibber's reappointment, for example), have been covered in previous editions of The Signpost. However, the report also gave details on a number of other projects not covered. For example, it highlighted the publication of a product whitepaper by the strategic product team (and the associated update from Sue Gardner) that will guide future engineering efforts, and the release this week, as expected, of version 1.0 of the new UploadWizard (see previous Signpost coverage).
Also in the report was the first target completion date of the new Virginia data centre (May 2011), and an update on Ryan Lane's efforts to create a virtualisation cluster, now expected sometime this month (see also an update on his blog). Russ Nelson's work on improving the media storage architecture is likewise progressing well; the necessity of this work had been highlighted by recent image thumbnailing failures (see this week's In brief). Work was also done to allow official surveys to pull data about users automatically, in order to shorten the amount of time required to complete them. The Foundation added that a beta release of version 1.17 of the MediaWiki software to external sites is expected "in early April", and a second deployment of the PoolCounter extension, first released last month (as covered by The Signpost) but quickly withdrawn after performance fears. A new project, the integration of a new caching system, known as EHcache, in order to decrease the number of requests that have to be handled directly, was also announced; in a less serious (but perhaps as important) development, the report noted that Ryan Kaldari had written a script to allow users to exchange gifts (including "virtual kittens") and other niceties more easily.
Readers interested in keeping up-to-date with specific projects may also be interested to note that the Foundation is trying harder to allow users to find this information and keep it up-to-date, for the benefit of staff, volunteer developers and users. This includes simplifying and improving the current system of Wikimedia blogs and creating a new page on the wikitech wiki to track recent and upcoming software changes, besides the server admin log.
Wikimedia's Google Summer of Code open for business
This year's MediaWiki Google Summer of Code (GSoC) scheme was announced this week on the Wikimedia Techblog. The blogpost, which served as a call for students and mentors for this year's programme, which is worth up to 5000 USD for budding students.
Over time, MediaWiki has benefited from GSoC students and their projects. For example, Samuel Lampa’s 2010 RDF import/export extension in Semantic MediaWiki is in use. And Jeroen De Dauw, GSoC student in 2009 and 2010, is now a persistently contributing member of the MediaWiki community, as is Brian Wolff, 2010 GSoC student...
This year’s ideas include writing and implementing cite templates in a PHP extension, improving the ImageTagging extension, XML dump work, pre-commit checks in our code repositories, and more. And of course we want to hear your own ideas, too! Interested? University, community college, and graduate students around the world are eligible to apply to Google Summer of Code. You don’t need to be a computer science or IT major, and you can work from home, [but you do need to] already know PHP. It’s also great if you have some experience with LAMP, MAMP, LAPP, or one of those kinds of stacks, and with the Subversion version control system.
If you’d like to participate, check out the timeline. Make sure you are available full-time from 23 May till 22 August this summer, and have a little free time from 25 April till 23 May for ramp-up.
The post has already attracted a number of proposals in the wikitech-l mailing list and elsewhere.
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 Foundation's image thumbnailing infrastructure has taken a battering over the last seven days, as the problems of scaling up the process to handle thousands of requests a day manifested themselves in slowness and frequent crashes. Fortunately, little of the damage was visible to users. Although there are few short-term fixes available to the operations team beyond redistributing the load among more servers, a long-term solution is in the pipeline (Wikimedia Techblog).
Administrators will now have to confirm that they actually intended to block themselves (revision #85025).
The release of version 2.0 of the article feedback tool was described on the Wikimedia Techblog. This month's report hinted that it could evolve into a more general review tool.
In a separate announcement from that quoted above, developer Mark Hershberger noted that PostgresSQL support was no longer considered vital for MediaWiki 1.17 and would instead be added in 1.17.1. This improves the chances of a quick release candidate for 1.17 (wikitech-l mailing list).