We had a translations issue on October 26th, 2017, and had to do a bit of scripting to get everything working.

When we scripted, apparently we changed the encoding (which I'm not 100% sure how this happened, as we copied the strings from TAC), and then when we've reimported them, they've stayed in the wrong encoding.

I'd consider this a bug, as there should be a pre-database check, and if it's incorrect, I think tac should either throw some kind of warning or (preferably) convert the text to the right encoding.

Not knowing that we have issues on our translations until customers report them is not good

https://jira.atlassian.com/browse/CONFSERVER-54554

https://jira.atlassian.com/browse/CONFSERVER-54348

You asked on the tickets how we translate - we normally don't have to touch them, it's that these were sent to gengo, and they came back invalid, so we had to run scripts on the error messages to fix them up (i.e. we got

). This is not normal, and not a part of our normal translations process. Really, if there could be sanitisation in the import process as well, that would save us a bunch of headaches  We have a lot of scripts to clean up language packs because we keep running into these problems, so it might be something we should pair on (we being the confluence and tac teams) to stop this kind of stuff from happening