Wikipedia:Wikipedia Signpost/2011-09-05/Technology report

Google Summer of Code: pencils down please
Work concluded this week on this year's MediaWiki-related Google Summer of Code projects. Of the seven projects that made it to the half-way stage, all passed their end of term assessments and will now submit their work to Google for auditing. (Google will then issue a final monetary reward to contributors for their volunteer development efforts over the summer.)

Of the seven projects, then, a number are likely to have an impact on Wikimedia wikis, including student Salvatore Ingala's project to make gadgets more easily customisable. One of his two mentors, Max Semenik, stressed that the project had been designed to make it compatible with the Foundation's own work to improve the usability of gadgets, and so it was unlikely to be discarded. He was also pleased with his student's progress in general, adding that constant intervention to keep the project on track and up-to-standard had not been necessary, with advice focussed only on small bursts at the beginning, middle, and end of the timeline. Yuvi Panda's attempt to make the compilation of large article subsets easier via a new extension also seems to fill a need onwiki. Other potentially Wikimedia-related projects include Kevin Brown's experimental ArchiveLinks extension and Aigerim Karabekova's work on Extension management. Also of interest is Akshay Agarwal's progress on separating the logic of logging in and registration from its presentation, with a view to allowing new ways to log in in future.

Other projects included work on the Semantic MediaWiki variant and on a Facebook-esque "status update" feature.

August Engineering Report published
The Wikimedia Foundation's Engineering Report for August was published last week on the Wikimedia Techblog and on the MediaWiki wiki, giving an overview of all Foundation-sponsored technical operations in that month (because of the discrepancy of Wikimania, the August report was published approximately two weeks after July's). Many of the projects mentioned have been covered in The Signpost, including the Wikimania and Developer Days, progress on HTTPS support, major work on customized campaigns for the Wiki Loves Monuments event, and the increasing readiness of both MediaWiki 1.18 and a new mobile platform for Wikipedia.

Nonetheless, the report also contained developments, which, although classified as major, have not yet received external coverage. In particular, a new team was set up within Wikimedia engineering to support a renewed effort at making Wikimedia-related software both readable and in particular more writeable in other languages (localisation and internationalisation). Comprising four staff members, its primary aim is to provide "a set of tools to facilitate editing in languages using a non-Roman alphabet". Also of note was a significant operations team meeting, focussed on improving its workflow and re-prioritising outstanding projects. Other projects which saw progress include data dumps (improved ability to restart failed processes without having to throw away previous progress) and the visual editor project, where the possibility of having a transaction-based edit system is being looked at. Such a system would allow for edits to be more easily merged in the event of edit conflicts.

Also included in the report was news that, after new users began to use the WikiLove extension as an easy way to send messages to other users, a new tool will be developed to provide a more consistent framework for this action. Meanwhile, developer Neil Kandalgaonkar "continued to work on real-time collaboration and is close to presenting a [working] demo" while Ian Baker "investigated and started to work on a chat system to be integrated to the concurrent editing interface, for collaboration and live help". Another project to look out for in the future is a renewed effort to allow a safe subset of arbitrary code to be included in pages to simplify template syntax.

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.


 * How to hold a Hackathon: The Foundation's Volunteer Development Coordinator Sumana Harihareswara posted a detailed list of things to consider when organising a Hackathon (a Wikipedia meetup with a strong technical focus). Hackathons are currently scheduled for the American city of New Orleans (October) and the British seaside resort of Brighton (November).
 * Improved HTTPS support comes to Wikimedia Commons: A series of bugfixes prompted a test deployment of improved HTTPS support to Wikimedia Commons and the WMF's own wiki (wikitech-l). The move establishes  as a second, (albeit still buggy) secure alternative to the   address. By incorporating developments regarding protocol relative links, the new site should provide a more integrated and secure browsing experiencing than the existing   site when fully complete.
 * New full-time member of staff: Aaron Schulz, a long-time developer of the MediaWiki software that powers Wikimedia wikis, has been hired full-time by the Wikimedia Foundation. As a part-time programmer, Schulz has worked on the flagged revisions extension, code review, and the heterogeneous deployment system that will allow MediaWiki 1.18 to be trialled on smaller wikis before it hits larger ones (wikitech-l mailing list).
 * Bugs fixed: approximately 70 bugs were closed in the last week, resulting in users of the (current) secure server no longer receiving non-functional "confirm email" links (bug #30647) and the gender label "unspecified" being changed to "undisclosed" to recognise that editors consider it an active rather than purely passive option (bug #30455).
 * Bug triage "pretty good": The latest bug triage, overseen by bugmeister Mark Hershberger, was held this week, focussing on internationalisation (I18N) related bugs (wikitech-l). "We focused on a small subset of the over 100 bugs tagged with 'i18n' and still managed to go over the allocated 1 hr for discussing things" wrote Hershberger (full notes from the triage are also available). Triages focussing on bugs related to the UploadWizard and the Wikibooks projects have been scheduled for the next two weeks (full schedule).
 * Advice to skin developers: Developer Daniel Friesen published his advice on writing MediaWiki skins compatible with the latest versions of the software on the wikitech-l mailing list. Skins are often used by third-party installers of wiki software to brand their local installations without touching the underlying software itself; in addition, registered users on Wikimedia wikis have the ability to choose their own skin from a limited selection.
 * [[Image:Cryptographic Hash Function.svg|thumb|300px|right|An example cryptographic hash function of the sort that may soon be included for revision text. Note that even small changes in the source input (here in the word "over") drastically change the resulting output, implying the strong link between "identical text" and "identical" hash which some argue would be very useful for researchers.]] Provision of revision text hashes: There was a discussion on the wikitech-l mailing list regarding the possible addition of a column of hashes to the revision table (now a full RFC). Although such decisions to undertake schema changes are not taken lightly, advocates of the addition argue that it would make reversions much easier to spot without looking directly at page text, as well as improving the ability for database dumps to have their integrity checked. Hashes are already provided in the image table on integrity-checking grounds.
 * Action pages to go?: A formal Request for Comment (RFC) has been opened on the issue of action pages (URLs of the form ), suggesting that they should become redirects to a uniform system of special pages (which are already in use for some actions such as blocking users). http://www.gossamer-threads.com/lists/wiki/wikitech/247348
 * Orphan image tagging to begin: Fbot (task 5), which will tag orphaned images to encourage their use, was approved this week. Discussions surrounding other bot tasks are continuing; outside input is vital if they are to be properly assessed.