TC based nightlies should have en-US and "multi locale" routes as well

RESOLVED FIXED in mozilla54

Status

Taskcluster
Task Configuration
RESOLVED FIXED
9 months ago
9 months ago

People

(Reporter: Callek, Assigned: Callek)

Tracking

unspecified
mozilla54

Details

MozReview Requests

Submitter Diff Changes Open Issues Last Updated
Loading...
Error loading review requests:

Attachments

(2 attachments)

(Assignee)

Description

9 months ago
Created attachment 8828457 [details] [diff] [review]
Diff of routes

Johan's work with automatically pushing aurora to play store depends on `multi` locale v2 routes. This patch adds some missing routes.

f? to mshal on the route diff

Initial attachment was generated with running full taskgraph --json on a before and after of the to-be-attached reviewboard patch.

and then running ::

 cat ../jobs_test2.json | jq '[. | to_entries[] | select(.key) | .value = .value.task.routes ] | from_entries' > ../routes2.json
 cat ../jobs_test1.json | jq '[. | to_entries[] | select(.key) | .value = .value.task.routes ] | from_entries' > ../routes1.json
 diff -U8 ../routes{1,2}.json
Attachment #8828457 - Flags: feedback?(mshal)
(Assignee)

Updated

9 months ago
Assignee: nobody → bugspam.Callek
Comment hidden (mozreview-request)

Comment 2

9 months ago
Comment on attachment 8828457 [details] [diff] [review]
Diff of routes

>callek@Centaurus:~/mozilla/hg/mozilla-central$ diff -U8 ../routes{1,2}.json 
>--- ../routes1.json	2017-01-19 14:26:56.068267633 -0500
>+++ ../routes2.json	2017-01-19 14:26:49.392254884 -0500
>@@ -1524,16 +1524,22 @@
>     "tc-treeherder.v2.date.10b21d912f60b8de5c655799a27e9229c8761bb1.0",
>     "tc-treeherder-stage.v2.date.10b21d912f60b8de5c655799a27e9229c8761bb1.0"
>   ],
>   "build-android-api-15-nightly/opt": [
>     "index.gecko.v2.date.nightly.latest.mobile.android-api-15-opt",
>     "index.gecko.v2.date.nightly.2016.11.06.revision.10b21d912f60b8de5c655799a27e9229c8761bb1.mobile.android-api-15-opt",
>     "index.gecko.v2.date.nightly.2016.11.06.latest.mobile.android-api-15-opt",
>     "index.gecko.v2.date.nightly.revision.10b21d912f60b8de5c655799a27e9229c8761bb1.mobile.android-api-15-opt",
>+    "index.gecko.v2.date.revision.10b21d912f60b8de5c655799a27e9229c8761bb1.mobile-l10n.android-api-15-opt.en-US",
>+    "index.gecko.v2.date.pushdate.2016.11.06.20161106160309.mobile-l10n.android-api-15-opt.en-US",
>+    "index.gecko.v2.date.latest.mobile-l10n.android-api-15-opt.en-US",
>+    "index.gecko.v2.date.revision.10b21d912f60b8de5c655799a27e9229c8761bb1.mobile-l10n.android-api-15-opt.multi",
>+    "index.gecko.v2.date.pushdate.2016.11.06.20161106160309.mobile-l10n.android-api-15-opt.multi",
>+    "index.gecko.v2.date.latest.mobile-l10n.android-api-15-opt.multi",

Just wanted to double-check, but since we generate en-US and multi-l10n as part of the same build, both of these artifacts will be present under both routes, correct? Kinda similar to how the l10n chunking ends up with several routes pointing to the same group of artifacts?

>+    "index.gecko.v2.date.revision.10b21d912f60b8de5c655799a27e9229c8761bb1.mobile-l10n.android-x86-opt.en-US",
>+    "index.gecko.v2.date.pushdate.2016.11.06.20161106160309.mobile-l10n.android-x86-opt.en-US",
>+    "index.gecko.v2.date.latest.mobile-l10n.android-x86-opt.en-US",

We don't currently publish an en-US route for android-x86: https://dxr.mozilla.org/mozilla-central/rev/aa3e49299a3aa5cb0db570532e3df9e75d30c2d1/testing/mozharness/configs/builds/releng_sub_android_configs/64_x86.py#4

I believe this is because we don't do mobile-l10n.py for android-x86, and thus don't have the other locales around that would make distinguishing en-US useful.

However, if you need this for consistency with the other android platforms to support some other bug, I don't think it is wrong to add it. It just hasn't been (as of yet) necessary.
Attachment #8828457 - Flags: feedback?(mshal) → feedback+

Comment 3

9 months ago
mozreview-review
Comment on attachment 8828459 [details]
Bug 1332407 - TC based nightlies should have en-US and "multi locale" routes as well.

https://reviewboard.mozilla.org/r/105858/#review107290

::: taskcluster/taskgraph/transforms/task.py:678
(Diff revision 1)
> +    task = add_l10n_index_routes(config, task, override_locale="multi")
>      return task
>  
>  
>  @index_builder('l10n')
> -def add_l10n_index_routes(config, task):
> +def add_l10n_index_routes(config, task, override_locale=None):

maybe call this "force_locale"?
Attachment #8828459 - Flags: review?(dustin) → review+
(Assignee)

Comment 4

9 months ago
(In reply to Michael Shal [:mshal] from comment #2)
> Comment on attachment 8828457 [details] [diff] [review]
> Diff of routes
> 
> Just wanted to double-check, but since we generate en-US and multi-l10n as
> part of the same build, both of these artifacts will be present under both
> routes, correct? Kinda similar to how the l10n chunking ends up with several
> routes pointing to the same group of artifacts?
> 

That is correct.

> >+    "index.gecko.v2.date.revision.10b21d912f60b8de5c655799a27e9229c8761bb1.mobile-l10n.android-x86-opt.en-US",
> >+    "index.gecko.v2.date.pushdate.2016.11.06.20161106160309.mobile-l10n.android-x86-opt.en-US",
> >+    "index.gecko.v2.date.latest.mobile-l10n.android-x86-opt.en-US",
> 
> We don't currently publish an en-US route for android-x86:
> https://dxr.mozilla.org/mozilla-central/rev/
> aa3e49299a3aa5cb0db570532e3df9e75d30c2d1/testing/mozharness/configs/builds/
> releng_sub_android_configs/64_x86.py#4
> 
> I believe this is because we don't do mobile-l10n.py for android-x86, and
> thus don't have the other locales around that would make distinguishing
> en-US useful.
> 
> However, if you need this for consistency with the other android platforms
> to support some other bug, I don't think it is wrong to add it. It just
> hasn't been (as of yet) necessary.

I don't *need* this for consistency, but it certainly helps keep the logic here simplified and free of extra hardcodes/forks. I don't think en-US as a unique thing is that useful, but I also don't see how the consistency in terms of the routes is a bad thing either.

Hope That Helps.
(Assignee)

Comment 5

9 months ago
mozreview-review-reply
Comment on attachment 8828459 [details]
Bug 1332407 - TC based nightlies should have en-US and "multi locale" routes as well.

https://reviewboard.mozilla.org/r/105858/#review107290

> maybe call this "force_locale"?

I agree, this reads better.
Comment hidden (mozreview-request)

Comment 7

9 months ago
Pushed by Callek@gmail.com:
https://hg.mozilla.org/integration/autoland/rev/4c3bda4c62e3
TC based nightlies should have en-US and "multi locale" routes as well. r=dustin

Comment 8

9 months ago
bugherder
https://hg.mozilla.org/mozilla-central/rev/4c3bda4c62e3
Status: NEW → RESOLVED
Last Resolved: 9 months ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla54
(Assignee)

Comment 9

9 months ago
Comment on attachment 8828459 [details]
Bug 1332407 - TC based nightlies should have en-US and "multi locale" routes as well.

This should land on aurora, its a regression of exposed routes between buildbot based nightlies and taskcluster based nightlies.

It has baked on central for around a week.
Attachment #8828459 - Flags: approval-mozilla-aurora?
Comment on attachment 8828459 [details]
Bug 1332407 - TC based nightlies should have en-US and "multi locale" routes as well.

Fix for android locale upload issues. OK for aurora!
Attachment #8828459 - Flags: approval-mozilla-aurora? → approval-mozilla-aurora+

Comment 11

9 months ago
bugherderuplift
https://hg.mozilla.org/releases/mozilla-aurora/rev/a943b51dbee9
You need to log in before you can comment on or make changes to this bug.