Closed Bug 749266 Opened 14 years ago Closed 5 years ago

Streamline deployment of l10n artifacts to preview, stage, and production

Categories

(Cloud Services :: Server: Identity, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WONTFIX

People

(Reporter: ozten, Unassigned)

Details

(Whiteboard: [qa+])

Our L10n process delivers the list of production locals while we're certifying our rpms that are destined to ship to production. locale strings and l10n-prod config are part of an rpm that include Node code. Any time this rpm is rebuilt, QA has to dedicate additional verification time. Can we re-engineer the deployment process for .po, .mo, .json language files, and a partial config for l10n-prod.json in a way that is 0 risk from QA's perspective? Starting point - can we isolate these files into a browserid-l10n.rpm file?
Whiteboard: [qa+]
Starting the devops work, I noticed... browser-server rpm still has .po files in it... I was under the impression that this work was done. I lost track of this original bugzilla bug, which indicates we still need to do this work. Goal: CI delivers server-browserid rpm and l10n-browserid rpm At a later time, we can update the l10n-browserid rpm with the latest strings from the community, right before pushing. This is lower risk since only strings are in l10n-browserid rpm. We can rework the proposed design, but the old model of testing an rpm and then cutting a new one 48 hours before deployment to production is what we want to avoid.
@lloyd - I've forgotten that we've removed node-gettext, can you comment on the build and runtime requirements of our current L10n solution?
is there documentation I can update that is out of date rather than typing into this bug report?
Via lloyd in IRC - .po files aren't needed at runtime

Closing as this ticket is for Persona

Status: NEW → RESOLVED
Closed: 5 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.