Translations

[simonsan]

Comprehensive Rust has developed a nice translation mechanism for mdbook books: https://github.com/google/comprehensive-rust/blob/main/TRANSLATIONS.md

We could make some research on how to adapt to that workflow making it possible to translate Rust patterns to different languages.

Newer simplified Chinese could be the starter.

[mgeisler]

Hi @simonsan, I would love to see the infrastructure used elsewhere. We will eventually publish the i18n helpers (google/comprehensive-rust#462), but for now I recommend installing them with

cargo install --git https://github.com/google/comprehensive-rust i18n-helpers

[simonsan]

Published now as https://github.com/google/mdbook-i18n-helpers 🚀

CC: rust-lang/mdBook#5 (comment)

[simonsan]

It will probably still take some time until we can fully adopt it, as it seems like that Markdown formatting isn't transferred in the translation process. Afaiu that could destroy the ability to further translate things within a language, when we would reformat the original markdown.

Tracking: google/mdbook-i18n-helpers#19

[mgeisler]

as it seems like that Markdown formatting isn't transferred in the translation process. Afaiu that could destroy the ability to further translate things within a language, when we would reformat the original markdown.

You are correct: the Markdown text is copied as-is to the PO files. So if you mass-format it, the translations will stop working.

My work arounds for this are:

• Implement Normalize Markdown in .pot files google/mdbook-i18n-helpers#19 that you linked above (harder)
• Implement a small tool which will apply formatting to the msgid entries in the PO files (much easier)

I hope to write a tool for the second option soon since I plan to standardize the formatting of my own Markdown files.

[simonsan]

Thanks for the clarification <3

[mgeisler]

Maybe I should clarify that the mdbook-i18n-helpers extract the text on a per-paragraph basis right now. So changes only invalidate a single paragraph at a time.

If you add/remove a few words in a paragraph, Gettext will do a good job at finding the old paragraph when you update your xx.po file for the xx language. The entry is then marked "fuzzy", which is a signal to the translator that they need to update the old translation.