Translations seem to be broken since at least one year (How to translate custom apps? - #9 by stephenBDT)
I am setting out to fix it but first would like to understand the general approach.
As I can see so far, the idea is, that there are two files, one with the source strings and one with the translated strings which map line by line (meaning soiurce string in line 15 in untranslated.txt will be translated with line 15 in translated.txt)
(assuming bench update-translations <lang> untranslated.txt translated.txt)
All this will result in <lang>.csv.
Thereafter, if there are new strings comming after this action, bench get-untranslated <lang> untranslated.txt should only contain new strings, that have not yet been translated in translations/<lang>.csv and when they are translated with a new translated.txt via bench update-translations <lang> untranslated.txt translated.txt, these new translations should be added to the translations/<lang>.csv file.
Is this roughly the idea?
What happens when we directly translate and/or modify the translations/<lang>.csv ? Is this something that can/should be done?
Please give me some informations, I am currently in the process of fixing the bugs and finishing up a pull request.
Thanks 