[Mailman-Users] accents in subjects

[Christopher P. Lindsey]

> >  However it's not only a question of 'Réf.' but of any
> > accentuated Subject: when it comes to the digest issue. So mailman is
> > also a little faulty here. 
> 
> Not really.  You cannot have 8 bit characters in header lines - the 
> character set etc applies to the body, so that subject is broken to 
> start with.

This is perfectly legal.  RFC 2047 supercedes RFC 822 and defines 
character encoding within headers, including the Subject: header.

> [from previous message]
> > Usually it get morphed from 
> >    Réf. : confirmation of subscription - request 170967
> > into something like
> >    =?iso-8859-1?Q?R=E9f._:_confirmation_of_subscription_--_request_170967?= 
> 
> so some MTA is converting the broken subject into some form of MIMEish 
> encapsulation - basically coping with a broken message by applying a 
> semi-broken workaround.

Regarding the use of Ref: vs. Re:, there's nothing in RFC 822 that 
requires it to be one way or another.  Subject: is defined as a 
*text type, which expands to "any CHAR, including bare CR & bare LF, but
NOT including CRLF".

> not sure you want mailman to start doing conversions around other 
> people's breakage.  One possibility might be to make the subject line 
> have more of a key in it hopefully picked to not have char set 
> implications - say that kast part be [[CONFIRM#170967]] and just look 
> for that style tag ignoring everything else in the line - making sure 
> that the tags chosen don't get hit by those 8 bit conversion problems.

I agree that this is a better solution.

Chris