Use it in the language that you prefer - if you spot issues, feel free to report them and we'll get them fixed. Ideally, you can edit the translation files yourself and submit a pull request, I have merged 3 such pull requests so far.

simakuutio, I was talking about the same thing in here, I think they are discrediting our work and saying they are "replacing it with AI", thing that is not true, as I watched 2 random languages, where there are little to no change since Martí had left (all I have saw are capitalization issues -at one language-, not actual translation issues). Atleast they should credit our work and not straight-up lie in our faces in the README.md file.
Update: they fixed this issue.

Luca (@lucadsign) I have restored the list of contributors for each language in TRANSLATION.md, and you are listed alongside Romanian: https://github.com/Devolutions/UniGetUI/blob/main/TRANSLATION.md

So what you're asking is that we end up with a two-tier translation system where we can flag translations as being made by AI as opposed to a human being, such that you, a translator who understands English, can configure UniGetUI to get partial translations with all the AI translations removed.

That ship has sailed, there's no easy way to track where the translations come from once the source of truth is git.

Also, preferring English language strings in a partial translation is only applicable to those that speak English as a second language. You are ignoring all users that do not understand English, and where such a fallback is not acceptable.

I'd rather have someone report incorrect translations that we can get fixed automatically than to have partial translations and users complaining about missing translations. If we were to follow the old, manual way, we'd be entirely reliant on volunteers to keep the translations updated, whereas now we can keep them at a 100% relatively easily.

So what you're asking is that we end up with a two-tier translation system where we can flag translations as being made by AI as opposed to a human being, such that you, a translator who understands English, can configure UniGetUI to get partial translations with all the AI translations removed.

That ship has sailed, there's no easy way to track where the translations come from once the source of truth is git.

Also, preferring English language strings in a partial translation is only applicable to those that speak English as a second language. You are ignoring all users that do not understand English, and where such a fallback is not acceptable.

I'd rather have someone report incorrect translations that we can get fixed automatically than to have partial translations and users complaining about missing translations. If we were to follow the old, manual way, we'd be entirely reliant on volunteers to keep the translations updated, whereas now we can keep them at a 100% relatively easily.

Is it important to be correct or just 100%? As I can see this is much important for you rather than having natives to translate the platform as it was. As a translator, reviewing my native languages, I found little to no mistakes when it was with Tolgee. I think that we should ditch AI and do it with people as it was and add credit into README.md. I think this is the best choice, because it's a higher chance that the translations will be better rather than slop. I don't think that it's hard to update the Tolgee changes and I think this is more efficient. I think we should stop using AI everywhere and think that human work has quality, not like AI. AI is not really good at translations, including my born (first native) language. There are a lot of mistakes that make no sense. In conclusion, I think that using only human translations is a better approach.

I understand that one change that made some of you angry were the modifications to the README.md. I'm working on splitting out that section to a TRANSLATION.md with the full list of languages, but also the list of contributors to each language, kept in sync from the Translators.json file. Would that meet you half way into making it more acceptable? I really do care about the 100% coverage, and even if you don't believe me, it's actually good for many users.

I understand that one change that made some of you angry were the modifications to the README.md. I'm working on splitting out that section to a TRANSLATION.md with the full list of languages, but also the list of contributors to each language, kept in sync from the Translators.json file. Would that meet you half way into making it more acceptable? I really do care about the 100% coverage, and even if you don't believe me, it's actually good for many users.

I understand that you do care, but that is not a quality idea. Maybe if the people using the app were more motivated, we could fix the thing that the translated languages are not 100%. And I think that pull requests are not a good idea because not many people know how to work with GitHub and Tolgee had a really great and inuitive UI. Really, I don't think that it's really hard to update the JSONs from answers that Tolgee had got but it will be better in the future for everyone, because it also has like memory for translation and it shows you the version and the percentage that it thinks it got right and you can compare as you are reviewing, it saved me a lot of time of writing because some of them were actually correct. Right now, I don't really thing they are that motivated with pull requests because it's not handy for most people. It's good to hear that you are crediting our work in translating this project.

That ship has sailed, there's no easy way to track where the translations come from once the source of truth is git.

I honestly thought of a quite simple system: the human translations are one file, the AI translations are another, if the option "Fallback to English instead of using Ai-generated translations" is unchecked (default behavior), the human file is loaded on top of the AI (with the top-most string being prioritized), if it's checked, then only the human file gets loaded.

That way you don't even have to adjust the Ai translation workflow, because the AI will always translate all string and store that in its file. Humans don't have to care about the AI strings, and every AI-only string won't be in the human file and as such be easily recognizable as AI-only.

simakuutio I have restored the list of contributors per language in TRANSLATION.md, and you are listed alongside Finnish: https://github.com/Devolutions/UniGetUI/blob/main/TRANSLATION.md

Let me investigate, that obviously should not happen

Can you tell give me some example of strings that aren't translated in Finnish that you noticed? As you said, it looks fine in the JSON, so the problem must be somewhere else in how the strings get loaded in the UI.

Hopefully this helps...

Forgot to mention, I'm using new UI because I wanted to see if it contains any issues and I'm pretty sure, this issue is related to this UI...

I confirm the issue, it does seem to be a problem with the new UI. We'll investigate, thank you for reporting it

Thanks for reporting, we'll look into it