Re: WG Chair Nits & start of WG Last Call: draft-ietf-secsh-publickeyfile-06.txt

[Ben Harris <bjh21%bjh21.me.uk@localhost>]

On Thu, 24 Mar 2005, Joseph Galbraith wrote:

> Ben Harris wrote:
> > A random interesting problem I observe with the header continuation format:
> > there's no way to write a header whose value ends with a backslash.  I 
> > don't
> > think this is worth worrying about.
>
> I think you are right in that it isn't worth worrying about it;

Especially since it isn't true.  A trailing backslash can be encoded as 
two backslashes followed by two linefeeds.  Sorry for the false alarm.

Something that does occur to me, though, is that the range of characters 
allowed in header-tags and header-values should perhaps be limited a bit. 
I'd suggest requiring those in header-tags to be printable and not ' ' or 
':', which is what RFC 822 requires.  I'm not sure I know Unicode well 
enough to know what's sensible in header-values, but I feel that at least 
excluding U+0000 would make life easier for C implementations, and it's 
obviously necessary to exclude U+000A and U+000D.

--
Ben Harris