From: Russ Allbery (rra@stanford.edu)
Date: Mon Jul 08 2002 - 12:02:43 CDT
Charles Lindsey <chl@clw.cs.man.ac.uk> writes:
> Actually, Russ's words also cover the matter of transporting such
> characters, which is already covered elsewhere (relayers MUST either
> copy everything or drop it). Moreover, mentioning control characters
> might just encourage the mess sendmail has gotten itself into by
> dropping 0c80-9F.
My language is describing the actions of a *news reader* when presenting
the article to the user. It's pretty much impossible to end up with that
problem when you're a news reader presenting the article to a user because
you're by definition a leaf node. It's impossible for any corruption
introduced at that point to propagate in any sort of automated fashion.
A news reader does need to be careful not to display raw control
characters, whether you want to put that information into the standard or
not. I have no strong opinion over whether to include it; if included, it
does belong in a note, and I can buy the argument that it's bloat.
I do oppose any language that implies that the news reader should display
control characters verbatim, though, because that's just dangerous. If
the news reader is running on a platform where 0x80-0x9F are control
characters (unlikely), it *does* need to strip, encode, or otherwise not
display them verbatim.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>