Wikimedia Language engineering/Reports/2016/February

TriangleArrow-Left.svgJanuary 2016 Wikimedia Language engineering monthly report for February 2016 March 2016TriangleArrow-Right.svg

Monthly updates from WMF Editing - Language team

Content TranslationEdit

Updates from the ongoing development and deployment for Content Translation. (Newsletter)

Development updateEdit

  • Translation sections are validated against the abusefilter rules in target wiki as the user edits it. This is based on the high number of publishing failures since abusefilter at target wiki blocks it. This will allow translators to make the edits valid before publishing
  • The saving of translations are now granular. Autosave feature saves changed sections instead of whole document every time. This is also expected to make the saving and restoring of translations more reliable
  • The translation storage code was refactored. The storage of translations will migrate to cx_corpora table from cx_drafts table.
  • If the source article of translation changed drastically and we are not able to restore an old translation against it, we will reload the old version of source.
  • Fixed the case of CX related attributes leaked into published article when a paragraph is broken in to two parts by enter key. The enter key behavior made consistant across browsers. It now adds a br tag instead of creating new block level tags like div or p
  • Translation dashboard will now show basic user statistics- Total translations, translations in current month and a simple graph
  • Translators get notified about the availability of personalized suggestions after second published translation
  • Parallel corpora development completed. The scripts for creating dumps in json or TMX format is developed. Planning to placed the dumps in - may be on monthly basis
  • In translation dashboard, marking a suggestion Favorite or Discarded will have a small animation to indicate where the items are added or removed
  • The link tool is no longer showed in section titles
  • Special:CX page, when accessed by a user not enabled beta feature, will provide a way to acitivate the feature with one click. The confusing message "No such page exists" was not helpful
  • Also other misc. bug fixes.

Usage dataEdit

Details Numbers
Articles created (during the month) 7806
Articles created till end of month 59829
Number of new translators (during the month) 1161
Highest number of articles created by one user

(cumulative number till end of the month)

Highest number of translators for a Wikipedia

(cumulative number till end of the month)

2042 (es)
Total number of translators with only 1 published article

(cumulative number till end of the month)

Total number of translators with only 2 published articles

(cumulative number till end of the month)

Total number of translators with only 3 published articles

(cumulative number till end of the month)

Total number of translators with only 4 published articles

(cumulative number till end of the month)

Total number of translators with >= 5 and <= 20 published articles

(cumulative number till end of the month)

Total number of translators with >20 published articles

(cumulative number till end of the month)

Articles deleted (during the month) 510


Other projectsEdit

Updates from Translate, Universal Language Selector, MLEB and other projects.

Development updateEdit

No MLEB was released this month. Next MLEB is planned for April.

February was the month of many cleanups and start of catching up with code review backlog. Translate went down from 35 open patches to 19 (with only 6 abandoned) and Universal Language Selector from 10 to 6. Both extensions have approximately one third of patches still waiting for explicit review.

Notable changes in Translate:

  • Željko Filipin submitted many clean-ups to the browser tests which haven't been maintained for some time.
  • A fix by Niklas Laxström was merged for the graphical glitches common on Special:Translate (T111685).
  • All messages now have message documentation.
  • JsonFFS now supported nested object structures with an implementation that can be shared with other file formats.
  • Nemo bis contributed patches to hide translator sandbox if not enabled, to allow aliases for message group ids to support renames and to automatically blacklist source languages for translatable pages.
  • The save button in proofreading more can now be used to mark the message as reviewed, patch by harshkothari410.
  • Ori Livneh compressed PNGs with zopflipng to squeeze some more bytes of the file sizes.
  • The script processMessageChanges.php got new options --safe-import and --name that help to further automate message processing and avoiding conflicts with multiple users respectively, patch by Niklas Laxström.

Babel and CleanChanges extensions were modernized by converting to extension registration, removing some deprecated code and adding standard code quality checks. LocalisationUpdate was also converted to extension registration and had some deprecated code removed.

An issue in TranslationNotifications that prevented using the "in other wiki" option was reported by MarcoAurelio and fixed by Niklas Laxström.

Usage dataEdit

  • Translatable pages: Meta 2657 (+45); 2961 (+54); Commons 437 (+1)
  • Translate extension is installed on 217 wikis (+8), according to Wikiapiary

Deployments and other site related updatesEdit

  • cxserver is migrated to scb cluster (Jessie) to use Node.js 4.2 with help of Tech Ops team.
  • MT:
    • Yandex MT service is available for Persian translations on Content Translation.
    • Apertium: Logs are now available at /srv/log as other services's log.
  • Campaign:
    • 'specialcx' campaign is enabled to all Wikis. This is easy way to go to Special:ContentTranslation page when user is logged out.

Cross team work/requirementsEdit

Team statusEdit

  • No change