Weblate Improvements
Background
The R Project has a global and active community with members speaking different languages around the world, in many cases with the need to be able to use R in their native natural language instead of English.
This project aims to provide better coverage of translations by making it easier to contribute translations to R – even without technical skills.
Problem statement
The R project has been supporting the translation of user messages, warnings, errors etc since R version 2.1.0 (April 2005) using
- the
gettextfunctions to lookup translations, and - the PO files defining the translations, stored in the SVN repository.
Although the technology existed, but contributing translations was not straightforward due to
- the technical overhead of dealing with PO files made it difficult for external parties to contribute,
- the limited write-access to the repository requires a Core R member to sponsor and actually commit the changes.
Around 2010, there was a web-based translation service hosted at http://translation.r-project.org/pootle/, which made the PO file strings available for translation using any web-browser after authentication. Unfortunately, the service was shut down later without a successor.
Proposed solution
A modern alternative of Pootle is Weblate, which is a web-based tool to make it easy to crowd-source translating strings extracted from the source code, written in Python, extensible by using plugins, comes with pre-build Docker images, great docs and community.
A POC installation has been deployed at https://translate.rx.studio in May 2022, and shared with the R community on the R Contributors Slack channel.
Early feedback was positive, and thousands of strings have been already translated in a dozen of languages, but there are ways to improve
- UX for translators and translation team (e.g. review workflows),
- reporting on the state of translations (e.g. GSoC dashboard),
- integration with R Core (e.g. quarterly patch files),
- maintenance (e.g. git conflicts and forced rebase),
- hosting (e.g. to decide if we keep the current domain or move under r-project.org, who is hosting etc).
Actual tasks are listed at the Weblate ideation dashboard.
Project requirements
Any help is highly appreciated, e.g. translations without any technical skills, but to achieve the goal of improving the current Weblate service, the below skills/experiences would be helpful:
- UNIX system administration (e.g. SSH, advanced git, Docker, AWS)
- Python (e.g. to write custom Weblate plugins)
- web services (e.g. to write an API endpoint in R that Weblate can call)
- project/product management (e.g. to define and document translation workflows)
Project resources
Project outcomes
- Recommendations and documentation for translation teams (e.g. workflows)
- Automated checks on translation suggestions
- More seamless integration with R Core (e.g. better docs and/or automatioin on generating patch files to be committed into SVN)