Putting this up for review mainly because I'm not sure whether there is any infrastructure to trigger or there are any other files to update after this change.

Hi,

I've added German people from translation because I can't tell if it's good or not.

Thank you,

Johnny

Thanks for going through the list. Didn't know about the wordlist on the website, sorry for that. Before I'll update the patch, let's discuss three points:

1. Coming from the repo, the wordlist is not that easy to find (at least for me, you may want to try it yourself). Only looking at the images, translation/recording contributors could miss e.g. if an adjective is meant. We should add this information inline (<lang>.json, words.json, filename, or somewhere else?) to provide context or at least add a direct link to the wordlist to src/activities/lang/resource. I wonder how the wordlist is generated in the first place (I count 54 "adjective"s, but nowhere near as much in the gcompris or gcompris-data repo). Please suggest a course of action, and I'll gladly help.

2. Please clarify if the translation should match the recording or the wordlist. I'm not too familiar with GCompris, but if you are shipping the translations and the recordings, it may be better if those matched (if the recordings are redone to match the wordlist, the translations could be changed again).

3. We should somehow mark the recordings which are currently wrong. Please suggest a place for this (Bugzilla, Wiki, task list, some status page, inline, ...?), and I'll add it there. (I guess we should also mark it as wrong when the recording says "ein/eine", so we could standardize on "der/die/das" instead.)

Let me know what you think and what I should do next, otherwise at least we fixed some typos :)

Hi,

In D7240#135648, @rkflx wrote:

Thanks for going through the list. Didn't know about the wordlist on the website, sorry for that. Before I'll update the patch, let's discuss three points:

1. Coming from the repo, the wordlist is not that easy to find (at least for me, you may want to try it yourself). Only looking at the images, translation/recording contributors could miss e.g. if an adjective is meant. We should add this information inline (<lang>.json, words.json, filename, or somewhere else?) to provide context or at least add a direct link to the wordlist to src/activities/lang/resource. I wonder how the wordlist is generated in the first place (I count 54 "adjective"s, but nowhere near as much in the gcompris or gcompris-data repo). Please suggest a course of action, and I'll gladly help.

To help finding the files, maybe we can add a text in the README to tell where to find the different files to translate and a link to the web page (http://gcompris.net/incoming/lang/words.html)?
The images and texts come from http://www.art4apps.org/images/

2. Please clarify if the translation should match the recording or the wordlist. I'm not too familiar with GCompris, but if you are shipping the translations and the recordings, it may be better if those matched (if the recordings are redone to match the wordlist, the translations could be changed again).

For me having the good text is better than having a good match so if there is a mistake we should fix it (and either remove or update the voice depending on the difference).

This part is tricky, because we don't ship the voices directly, we can update them independantly with the words (basically, we don't often fix the texts) and they are not linked to a specific version. If we fix a text or a voice, the user would have to update both to have the good match.
For me, it will be good to update the text and voices when we release.

3. We should somehow mark the recordings which are currently wrong. Please suggest a place for this (Bugzilla, Wiki, task list, some status page, inline, ...?), and I'll add it there. (I guess we should also mark it as wrong when the recording says "ein/eine", so we could standardize on "der/die/das" instead.)

For the introduction voices, we added a .po file and we asked for translators to only translate the ones that have voices, mark fuzzy the voices that are not good and don't translate the ones that don't have voices. But we don't use it directly in the code, it's only help for translators/people who record voices.
I'm guessing we could do the same (and adding a context to the translation) but we need to find how to load it dynamically in the code (or having a script that would synchronise them).

Let me know what you think and what I should do next, otherwise at least we fixed some typos :)