@@ -238,6 +238,25 @@ When assigned a Translated Documentation Push MR:
Note: Maintainers do not review language content or translation quality, as this occurs earlier in the translation workflow on forks and in the translation management system (Phrase).
### Syncing forks with upstream projects
Content syncing between upstream projects and downstream forks is automated using the [Automate tech docs fork](https://gitlab.com/gitlab-com/localization/tech-docs-forked-projects/automate-tech-docs-fork) project. The process has two independent steps: [Pull Mirroring](https://docs.gitlab.com/user/project/repository/mirror/) syncs each fork's `master`/`main` branch with its upstream counterpart, and a separate pipeline sync merges those changes from `master`/`main` down to the `main-translation` branch at 12-hour intervals. The two steps are deliberately decoupled so that the sync pipeline has no direct dependency on upstream projects (no bot memberships or access tokens in upstream).
Failure reports are posted automatically to the public [#localization-alerts](https://gitlab.enterprise.slack.com/archives/C095BS4NLE4) Slack channel. For troubleshooting mirroring failures, pipeline errors, and access token issues, see the [Automate tech docs fork project documentation](https://gitlab.com/gitlab-com/localization/tech-docs-forked-projects/automate-tech-docs-fork).
### Enabling the translated documentation site
To toggle the translated site in production:
1. Edit `config/_default/hugo.yaml`
2. Change the language setting:
```yaml
languages.ja-jp.disabled:false
```
This single configuration change activates or deactivates the translated documentation site for a given language.
## Translation workflow
We've established a two-tier translation workflow that balances speed and quality. AI-assisted translations are published to the documentation site before linguistic human review is complete, giving users faster access to the latest information. These translations are then reviewed by professional and specialized human translators and updated as needed for accuracy and quality. This approach ensures the localized documentation site stays as close as possible to the English version while we iteratively improve translation quality.
