Investigate moving localization off CurseForge #949
Labels
Comments
Meorawr
added a commit
that referenced
this issue
Jul 5, 2024
Following on from the lessons previously learned in #946 and #948, we'll set the delete-missing-phrases flag _only_ on enUS uploads going forward as that appears to work fine. All other translation uploads will not set the flag. This means that going forward the minimum required process to delete a translation string should just be to remove it from the 'enUS.lua' file; a subsequent import of translation strings would then remove it from all other files automatically. It is, however, totally permissible to remove the string from all files if that would make things clearer. Additionally I've made some of the terminology around the place a bit more consistent for this whole process, eg. preferring to use "translations" instead of "locales" for target names and workflows. For the possibility of using non-CF sources (#949) I've also removed "CurseForge" from the workflow strings. Finally - one functional change; we no longer import translations on commits to main. We continue to export new strings from the enUS file, but that's all. The importing of non-enUS translations continues to happen on the same timer it always has, as well as being manually runnable.
Meorawr
added a commit
that referenced
this issue
Jul 5, 2024
Following on from the lessons previously learned in #946 and #948, we'll set the delete-missing-phrases flag _only_ on enUS uploads going forward as that appears to work fine. All other translation uploads will not set the flag. This means that going forward the minimum required process to delete a translation string should just be to remove it from the 'enUS.lua' file; a subsequent import of translation strings would then remove it from all other files automatically. It is, however, totally permissible to remove the string from all files if that would make things clearer. Additionally I've made some of the terminology around the place a bit more consistent for this whole process, eg. preferring to use "translations" instead of "locales" for target names and workflows. For the possibility of using non-CF sources (#949) I've also removed "CurseForge" from the workflow strings. Finally - one functional change; we no longer import translations on commits to main. We continue to export new strings from the enUS file, but that's all. The importing of non-enUS translations continues to happen on the same timer it always has, as well as being manually runnable.
Solanya
pushed a commit
that referenced
this issue
Jul 5, 2024
Following on from the lessons previously learned in #946 and #948, we'll set the delete-missing-phrases flag _only_ on enUS uploads going forward as that appears to work fine. All other translation uploads will not set the flag. This means that going forward the minimum required process to delete a translation string should just be to remove it from the 'enUS.lua' file; a subsequent import of translation strings would then remove it from all other files automatically. It is, however, totally permissible to remove the string from all files if that would make things clearer. Additionally I've made some of the terminology around the place a bit more consistent for this whole process, eg. preferring to use "translations" instead of "locales" for target names and workflows. For the possibility of using non-CF sources (#949) I've also removed "CurseForge" from the workflow strings. Finally - one functional change; we no longer import translations on commits to main. We continue to export new strings from the enUS file, but that's all. The importing of non-enUS translations continues to happen on the same timer it always has, as well as being manually runnable.
|
Figured I'd mention it: Another website I see used a lot for open-source projects on GitHub, aside from Weblate, is called Crowdin. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The CurseForge localization system is absolutely dire and we're questioning if the supposed benefit of easy access is really worth it. Alternative more generic systems exist like Weblate - we could possibly look into using those instead.
One challenge of note; the CF import process accepts a Lua chunk and parses it for us. Any alternative won't know about Lua so we'll need some script support to parse our strings and put them into an intermediate format (.po, .ts, .csv, whatever is most convenient - the hard work is more in the parser).
The text was updated successfully, but these errors were encountered: