(#zloep2q) @movq@www.uninformativ.de If my memory serves me right, I think v2 doesn’t mention UTF-8 at all. Then I came along and noted that the Content-Type: text/plain might be not enough, as the HTTP spec defaults to Latin1 or whatever, not UTF-8. So there is a gap or room for incorrect interpretation. I could be wrong, but I understand @anth@a.9srv.net’s comment that he doesn’t want to even have a Content-Type header in the first place.

I reckon it should be optional, but when deciding to sending one, it should be Content-Type: text/plain; charset=utf-8. That also helps browsers pick up the right encoding right away without guessing wrong (basically always happens with Firefox here). That aids people who read raw feeds in browsers for debugging or what not. (I sometimes do that to decide if there is enough interesting content to follow the feed at hand.)