This does fix it in this case, but it feels wrong to first decode the password and then encode it again when we can just use the bytes that are already there, which is what I did in f3f5f8384b57531aee463dbab264b0d458561983. I frankly don't know why the code was so weird. The other advantage of f3f5f8384b57531aee463dbab264b0d458561983 is that it adds a regression test, which is why I will prefer it over this PR. I hope this isn't taken as any sort of offense, I really appreciate the contribution very much!
This does fix it in this case, but it feels wrong to first decode the password and then encode it again when we can just use the bytes that are already there, which is what I did in f3f5f8384b57531aee463dbab264b0d458561983. I frankly don't know why the code was so weird. The other advantage of f3f5f8384b57531aee463dbab264b0d458561983 is that it adds a regression test, which is why I will prefer it over this PR. I hope this isn't taken as any sort of offense, I really appreciate the contribution very much!