From MozillaWiki
Jump to: navigation, search


  • l10n next actions
    • axel and joduinn had some next actions to work out which locales are based on m-c and which are based on aurora.
  • l10n coverage
  • Metrics & stability report
  • Branch naming proposal
  • Abuse of tracking flag
  • Triage

l10n next actions

(I captured this from a summary that joduinn sent out on Friday)

  • locales that use mozilla-central as source-of-truth
    • do l10n nightly builds, and provide nightly updates
    • localizers migrate to aurora every 6 weeks - on same schedule as developers (every localizer migrates their own locale)
    • on aurora, do l10n nightly builds, and provide nightly updates
    • Ehsan wants "ar" to be one of these locales, to test RTL
    • Axel to create list of locales that
      • a) want to work on m-c (happy to work before string freeze)
      • b) have the required hg skills to accurately double land
      • c) locales that are important in specific global regions (ie china?)
      • d) where localized users prefer nightly to aurora
    • Current estimate ~20ish locales
  • locales that use aurora as source-of-truth
    • on m-c, do not generate l10n nightly builds, and provide nightly updates
    • do not migrate repo to aurora every 6 weeks
    • lock down repo on l10n-central as it is not being used
      • reduce confusion, risk of error
      • still allow "l10n-drivers" to edit (escape hatch for emergency)
    • localizer lands changes on aurora only
    • on aurora, do l10n nightly builds, and provide nightly updates
    • Axel to create list of locales that prefer to work once string freeze happens (ie migration to aurora)
    • Currently ~60ish locales
  • Next actions
    • Axel to provide list of early/experimental "stringfreeze/aurora" only locales
    • Axel to provide two lists (m-c locales, aurora locales)
    • joduinn to create new ldap group "l10n-drivers"
    • Once RelEng has both lists from Axel;
      • for any locale that wants to use aurora as source-of-truth
      • stop generating l10n nightlies on m-c by removing from all-locales in m-c
      • generate custom updates to migrate those l10n users to aurora, QA then live
      • this is not blocked on (workaround exist)
    • axel, joduinn to meet again next week about l10n dashboard
  • Want to revisit whether or not we use merges

Localizations coverage

We have some locales that we used to ship but that haven't been active for a while:

Mongolian (mn) 
no contact for the longest time, 2.5k on 3.0, 3.5k on 3.5. Updates to en-US?
Occitan (oc) 
localizer not happy with our process, 250 on 3.6, updates to en-US
Georgian (ka) 
no active, but loose contact to localizer, need to reanimate -- no need for beta or release builds
Khmer (km) 
maybe upcoming, but no need for beta or release builds (no ADUs says pentaho?)

Localizations that may be good candidates for focus on aurora: African locales mentored by Dwayne. Message TBD.

Upcoming Localizations:

Swahili (sw) 
to be built for just Aurora, sw hg.

Branch naming proposal

  • Need Christian to close on whether or not he is ok with what we have done.
  • We will follow-up on Aurora call on Thurs.

Metrics & stability

  • 6.0a2 currently had 27K users yesterday.
  • At it's peak, 5.0a2 had about 46K users, it currently has 23K.
  • The trunk peaked at about 50K users for 6.0a1.
  • Although we have made steady progress moving people over to Aurora, the transition to Beta which ramped up quickly to 1 million+ users demonstrated that we really need more people on that channel to help us find stability regressions sooner.
  • Just had a meeting to discuss how to grow the channel over time. We still have a problem where we lose users when we do an update and they don't migrate.
  • Didn't do the AMO bump, might be an issue but not really sure.
  • Need to get better data around why people are refusing the updates.
  • Asa - might be some benefit of users not getting nightly updates.

Abusing "tracking" flag like old "blocking" flag

  • Concern that some people are treating tracking as blocking
    • "This bug is important, I want pressure on it / it prioritized"
  • Christian will post on the channel blog trying to clarify