Closed Bug 1345340 Opened 8 years ago Closed 8 years ago

be locale still auto filled in new desktop beta l10n-csets but does not exist shipped locales

Categories

(Localization Infrastructure and Tools :: Administration / Setup, task, P1)

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: jlund, Assigned: Pike)

References

Details

The description of: "Bug 1304747 - Redirect existing Belarusian (be) users to Russian (ru) builds via update snippets" suggests we have dropped "be" but Elmo/ship-it still recommends it when submitting to shipit Should we still be doing "be" locale based betas for 53.0? FTR - we removed "be" from beta1.
(In reply to Jordan Lund (:jlund) from comment #0) > Should we still be doing "be" locale based betas for 53.0? flod ^ do you know what the state of "be" is?
Flags: needinfo?(francesco.lodolo)
Belarusian needs to stay on aurora for a few cycles, not ship Beta or Release. I started taking sign-offs on Elmo for my mental sanity, as in checking periodically what they commit (it's a brand new team) and stamp that it's OK. I never done that before, so we might not have it this bug so far (but I've seen Delphine doing it for Fennec, not sure if that's going to cause similar issues). @pike https://l10n.mozilla.org/shipping/l10n-changesets?av=fx53 This shows Belarusian, since it has a sign-off, but it's not in shipped-locales. Is it something that needs to be fixed on Elmo or ship-it?
Flags: needinfo?(francesco.lodolo) → needinfo?(l10n)
In the end, I think I'd like to solve this in a way similar to what we do on mobile now. Assuming that works as nicely in practice as the theory predicts. That is, we'll have something that takes the revisions from elmo, and the platform/shipping status from the repo, and lands the changesets in tree to be used by the builds. I'd like that to ignore revisions from elmo that it doesn't have in the in-tree data, basically acting as a filter. Probably good to report on locales in-tree but without revisions, but they should also be dropped from the generated file, I think. That is, IMHO, we shouldn't fix ship-it nor elmo ;-) If this is causing immediate trouble, the most straightforward way might be "not to", aka, I'd tweak the data to take the 'be' sign-off off the table again. I understand that flod would like to have this as a tool to get belarusian ready to ship, but then the in-tree configs might not be all that far away for desktop as well? Unless it's actually an easy fix on the ship-it side, or wherever the symptoms show up, no idea how hard that'd be.
Flags: needinfo?(l10n)
I have no visibility on how ship-it actually works, by looking at the code all interactions with elmo seems to happen at JS level, reading the JSON and filling a field https://github.com/mozilla-releng/ship-it/blob/master/kickoff/static/suggestions.js
Priority: -- → P1
I agree with using an auto-bumped l10n-changesets.json for desktop mid- to long-term, but it may take a while (days/weeks) to make happen. Do we have a reasonable short term solution?
We talked about this on IRC, and agreed that the easiest way to fix this short-term is to remove 'be' from the currently signed-off locales again. Aki's gonna file a bug to get l10n-bumper for desktop, too, and then we can work on pre-release locale sign-offs incrementally again. Until then, flod agreed to work with Belarusian without the sign-off marker until we either have bumper or Belarusian is intended to ride the trains to release. I've modified the elmo database accordingly, so we can resolve this bug FIXED.
Assignee: nobody → l10n
Status: NEW → RESOLVED
Closed: 8 years ago
Component: Ship It → Administration / Setup
Product: Release Engineering → Localization Infrastructure and Tools
QA Contact: rail
Resolution: --- → FIXED
Blocks: 1345619
You need to log in before you can comment on or make changes to this bug.