APM trace screen is blank

[mirai-jeff]

Kibana version:
v7.8.0

Elasticsearch version:
v7.8.0

Browser version:
Chrome Version: 84.0.4147.89（Official Build） （64 bit）

Browser OS version:
macOS Mojave 10.14.6

Original install method (e.g. download page, yum, from source, etc.):
Elastic cloud

Describe the bug:
Adding i18n.locale: "ja-JP" to the kibana configuration will cause the APM trace to go blank.

The following Javascript error has occurred:

INFO: 2020-07-22T05:18:04Z
  Adding connection to https://xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.ap-northeast-1.aws.found.io:9243/elasticsearch
kbn-ui-shared-deps.js:342 Error: [I18n] Error formatting message: "xpack.apm.percentOfParent" for locale: "ja-jp".
TypeError: Cannot read property 'length' of undefined
    at Object.t.translate (kbn-ui-shared-deps.js:312)
    at PercentOfParent (6.plugin.js:1)
    at ua (kbn-ui-shared-deps.js:342)
    at xs (kbn-ui-shared-deps.js:342)
    at ml (kbn-ui-shared-deps.js:342)
    at hl (kbn-ui-shared-deps.js:342)
    at ol (kbn-ui-shared-deps.js:342)
    at kbn-ui-shared-deps.js:342
    at t.unstable_runWithPriority (kbn-ui-shared-deps.js:350)
    at Hi (kbn-ui-shared-deps.js:342)

Steps to reproduce:

1. View APM user trace page.

Expected behavior:
The content is displayed normally.

Screenshots (if relevant):

Errors in browser console (if relevant):

Provide logs and/or server output (if relevant):

Any additional context:

[elasticmachine]

Pinging @elastic/apm-ui (Team:apm)

[smith]

A similar error happens in logs with xpack.infra.logs.customizeLogs.textSizeRadioGroup:

[smith]

@mirai-jeff #72984 should fix this. It was due to incorrect formatting of one the translation files. They use ICU-style select statements which have some tricky formatting. My PR fixes the formatting but the changes I made probably aren't grammatically correct because I don't speak Japanese.

Any advice on my PR would be appreciated but we'll also have our translation team look it over. Thanks for reporting this.

[smith]

#67082 was a similar issue that affected 7.6 whose fix shipped in 7.8. We've had improvements to our internal translation tooling that better detect things like this. In #72984 I looked and found all the places where this issue was present, so this shouldn't come up again unless new translations are introduced that our tooling doesn't catch.