John Gardiner Myers wrote:
> If Mozilla detects that a raw 8-bit header conforms to UTF-8 syntax it
> will interpret the header as UTF-8, regardless of the folder charset.
> This semantic is not in Communicator.
I saw that in the source, it's nice for usefor.
Still, if for some reason, something that is not utf-8 happens to be a
non-illegal utf-8 string, it will interpreted as utf-8.
It can be accepted, as it's definitively not the most common case.
> Having Mozilla *generate* raw UTF-8 in headers is an entirely different
> question.
That's what it does now, but it's a bug.
There is a feature, required by many users, that is to send raw data in
local encoding.
This is completely incompatible with the sending of raw UTF-8 in headers as
there no systematic way to tell the difference between the two.
Unfortunately, many users want that in the current state of things, and allowing
to choose between the two behaviours, would be a lot more work in Mozilla than
just doing the correction.
