Re: Aging Dependencies?

>It's actually more complicated than that - it gets tied into the way
>exmh chooses 'faces' and fonts, and is handled differently for scan
>output and the message display.

Yeah, but I'm still trying to figure out what turned a particular single-byte
ISO-8859 character character into a 2-byte UTF-8 sequence.  It wasn't
nmh; was it Tcl?

In the case you describe below, it seems like people should be "encouraged"
to use a Unicode font and convert those headers to UTF-8 internally
irregardless of the user's locale.

>One corner case in particular is in
>msgShow.tcl, if m_tagnames specifies a highlighting for a header line
>that includes a font that doesn't match the encoding (for instance,
>specifying that From: is blue with a particular iso8859-1 font, but
>the header included a =?utf-8?q?whatever field), you end up with good
>and proper mojibake.

Speaking of mojibake ... the part of the message where you quoted
me was interesting.  The character set was iso-8859-1, the encoding
was quoted-printable.  Unicode codepoints in my message that are
valid in ISO-8859 such as U+00E4 were properly encoded using
quoted-printable.  But Unicode codepoints that were NOT valid in
iso-8859-1 (U+201C, U+201D, U+FFFD) were simply spit out as UTF-8
bytes and NOT encoded properly via quoted-printable (which as I
read it is a violation of RFC 2045).  I don't know if that's how
the email left your SMTP server (the mailing list can rewrite
messages).  If you don't mind me asking ... how did you compose your
reply message?  This is all part of "trying to understand how people
deal with character sets in the real world".

--Ken