Ok. So I should encode the data from utf-8 and decode it to iso?

No. Don't go mixing all the terms I've used.

You should decode incoming data (from UTF-8 or whatever encoding it is) into perl's internal format.

Then do your string operations with decoded strings.

Then when you ouput it, encode it. It's not the right format already - it's in Perl's internal format, which can be either latin1 or UTF-8, depending on some factors you shouldn't care about.

Please read this, it explains it all in sufficient detail (I hope).

I meant "decode from" and "encode to" - sorry for the mixup

The thing is that everything else that is printed (without setting the binmode) gets printed ok (does this mean that I already have the correct output mode?). So it feels like I'm trying to decode data that isn't utf-8 from the beginning? (Can I test the incoming data in a simple manner?) I will look into the link you provided.

(Can I test the incoming data in a simple manner?)

See the documentation for Encode::decode - you can tell it to die on invalid input.

Perl 6 - links to (nearly) everything that is Perl 6.