[Top] [All Lists]

FWS *within* <quoted-string> (was: rfc2821bis-08)

2008-02-25 20:21:20

John C Klensin wrote:

The "Last" in "Last Call" needs to mean something

"Your *last* chance wrt this document", rare cases
of appeals and repeated Last Calls not withstanding.

It is not an accident that HT is still prohibited
in addresses.

After RFC.klensin-net-utf8 was approved (14:04:33)
I'll consider this as fait accompli in 60 days.  

Now 2822upd has to get rid of it, it makes no sense
to forbid HT in 2821bis and to allow HT in 2822upd.

Ideally it is a semantical instead of a syntactical
problem, we could say that any FWS actually stands
for a single space.

But everybody expects <quoted-string> to mean what
is quoted *as is*.  The 2822upd theory is that it
means what it says modulo <quoted-pair>s, minus the
leading backslashes of <quoted-pair>.

In practice RFC 2617, RFC 2831, RFC.usefor-usefor,
and anything building on it use a <qdstr-val>, i.e.
keep also <quoted-pair> *as is*.

Alexey's 2831bis I-Ds proposed to "fix" <qdstr-val>,
but after nobody produced the time-machine for also
fixing RFC 2069 gave up on this idea.

And on the URI list somebody proposed years ago to
get rid of these weird backslashes at least in tag
URLs, but the tag + mailto + news-nntp URI authors
declined, IIRC.

IMO that reduces the 2822upd <quoted-pair> theory to
"maybe user agents displaying a <quoted-string> get
it right".  This issue is critical, it could affect
DKIM canonicalizations or mailto-bis or RFCs sitting
in the RFC editor queue.


<Prev in Thread] Current Thread [Next in Thread>