On 6 Feb 2003 michael@??? wrote:
> Exim 4.12 defines header fields in src/expand.c:445 as sequence of
> characters of
>
> mac_isgraph(*s) && *s != ':'
>
> Said macro is:
>
> #define mac_isgraph(c) \
> ((uschar)(c) > 32 && (uschar)(c) != 127)
>
> That includes in 8-bit characters.
>
> RFC 2822, section 3.6.8 says:
>
> field-name = 1*ftext
>
> ftext = %d33-57 / ; Any character except
> %d59-126 ; controls, SP, and
> ; ":".
>
> That forbids 8-bit characters. Shouldn't Exim follow the RFC there?
Exim is 8-bit clean and always has been. Along with some other MTA
writers, I feel that forcibly restricting messages to 7-bit characters
is silly nowadays. If you are going to allow 8-bit characters, I see no
point in allowing them only in some places and not others.
Every now and again there is an inconclusive discussion, bordering on a
flame-war, on certain mailing lists. There is very little progress
towards an agreement on extending the official character set, though I
think everyone agrees that it needs extending. The latest discussion
(calmed down now) was "just go for Unicode (UTF-8)" vs "you need
complicated encoding mechanisms to maintain compatibility".
Inconclusive, as I said.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
