ImprovingTranslationStatusReporting

Summary

The purpose of this specification is to provide better translation status reporting at the distro level in a resource external to Launchpad Translations.

The main goals are:

  • Ubuntu translators, users and customers have an overview of the overall status of translation of the distro for a given language
  • Translations which need work done are easily identified

Release Note

A summary of per-language translation status at the distro level is publicly available for Ubuntu. This includes the list of languages in which we consider the translation of Ubuntu as complete.

Rationale

Currently Launchpad Translations offers an overview of the translation status of all translatable packages. While complete, this includes a lot of packages which are either not part of the default distribution or which contain strings not visible to the regular user. These contribute to the overall statistics, giving an inaccurate view of the translation status at the distro level.

As the focus of the Launchpad Translations developers for the 4.0 cycle will be on upstream integration, there will be no new features in Launchpad in this regard, and thus better reporting at the distro level will have to be implemented externally.

There are several reasons why we want to provide better status reporting:

  • At present we have no clear quantitative translation goals for Ubuntu
  • The current reporting does not support working towards a clear goal nor making quantitative statements about a release, nor how language pack releases change the status
  • Providing a public summary of translation status per language would help driving Ubuntu adoption when users or OEMs are considering to use or to ship the distro in a particular language (or set of)

User stories

  • Pierre translates Ubuntu into French and would like to know in which translations he should focus and how his translation contribute to the overall distro translation. The available summary report helps him in that.
  • Mr. Bell is considering selling computers with Ubuntu preinstalled in Brazil and Italy. Knowing from the summary report that Ubuntu has good language support in Brazilian Portuguese and Italian makes his decision easier.

Design

  • Build upon the work done on the previous cycle to provide statistics at release time and provide scripts to list translations from the default installation which might need work.
  • Provide a web UI to display the statistics. This will probably initially be a set of simple HTML pages.
  • Identify the high priority packages.
  • Identify the high priority languages. A criteria might be the number of speakers or the number of users
  • Identify their current translation status.
  • Have plan to engage Ubuntu Translators to focus on the high priority packages and languages and how we can have a final translations push later in the cycle to get nearly complete translations up over the barrier in which we consider them complete.

Implementation

The following actions have been identified for the implementation of this specification:

  • [danilo] Add more information to the current raw data export -> (number of strings that needs review or changed from upstream, date of last change done for a template for each language, priority value)

  • [danilo] Create raw data export for Lucid

  • [dpm] Work with adiroiban to finalize the templates priority -> https://wiki.ubuntu.com/Translations/TemplatesPriority

  • [dpm] Identify high priority languages
  • [dpm] Provide scripts to identify high priority translations which might need work
  • [dpm] Provide a web UI to present a summary of the statistics
  • [dpm] Define language tiers and process for promoting languages.

Unresolved issues

At this point only statistics for Ubuntu will be provided. If time allows, status reports for Kubuntu and UNR will also be generated.

As we are working on raw statistics and thus cannot identify the procedence of the strings (e.g. gconf schema strings are not visible to regular users, and should have a much lower priority for translation than e.g. Nautilus menus), priorities will be assigned on a template (translation) level rather than on a message level. As a result, the report will not be as accurate as it could be, but it will nevertheless provide a good overview.

BoF agenda and discussion

UDS session notes: 

== Improving translation reporting ==

 * Kyle: organize both languages and packages into categories to express how 
   well Ubuntu is translated for the important languages and important languages
 * Definitions of language tiers must be tied to the activity and support of the
   language, not only statistics

 * make translation reports/status visible to users / decision makers/takers

== What to report on ==

Assuming we have the data (one way or another: manual data export or LP API), we should:

 * How well supported Ubuntu is in your language?
 * What needs work?
 * perceived completeness of translations / per language

To properly report on what the status is, we need to define language and package tiers.

Adi interested in reporting the deltas with upstream

== Actions ==

 * Danilo to add more information to the current raw data export:
   * number of strings that needs review or changed from upstream
   * date of last change done for a template for each language
   * priority value
   * create report for curent stable and dev release

 * Finalize templates priority (Adi/David)
   * https://wiki.ubuntu.com/Translations/TemplatesPriority

 * David to define language tiers and process for promoting languages.

CategorySpec

Specs/ImprovingTranslationStatusReporting (last edited 2009-12-02 13:15:57 by 212)