ietf-822
[Top] [All Lists]

Re: Attempts at establishing harmful convention with multilingual עברית /English subjects

2004-12-01 15:28:53

On Wed December 1 2004 13:30, Nathaniel Borenstein wrote:

But even then, what's an "intelligent" thing for software to do?  If 
we're a polyglot mailing list discussing complex linguistic issues, 
does the fact that my subject line is all Hebrew, Arabic, and French 
somehow imply that "Re:" is the wrong reply convention?

That's up to the author of the message; software should provide
the author with the ability to view metadata such as charset and
language information (and to provide the author with the ability
to edit that information from a set of legal values when
generating a message).

, or that any  
inserted [] prefix should occur in any particular language?

If such insertion is being performed by the message author, the
remarks above apply.  If you're taking about some other software
modifying an author's field without his knowledge or approval,
I've already said that I think that shouldn't be done.

I've said elsewhere (mail-ng) that there are a number of
reasons favoring splitting message content into clearly-delineated
areas where message authors, user agents, transport agents,
list expanders, etc. can each place relevant information without
getting in each other's way.  Having intermediaries screw around
with author's information hampers certain operations (e.g.
signing of message content including author header fields such
as Subject, To, From, Date, Keywords, Cc, Reply-To).  The only
way signing can work (by definition, since the purpose is to
verify that content hasn't changed) is if the signed information
remains unaltered.  And before you mention triple-wrapped
security multiparts, make sure that you  can point to UAs that
can generate and read triple-wrapped messages (and that
would include details such as displaying the Subject, etc.
fields from the signed part in message "folder" lists).