W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2008

Re: i74 proposal take 2

From: Frank Ellermann <nobody@xyzzy.claranet.de>
Date: Fri, 28 Mar 2008 04:08:05 +0100
To: ietf-http-wg@w3.org
Message-ID: <fshna2$9to$1@ger.gmane.org>

Mark Nottingham wrote:

> One question to consider here -- should %x80-%x9F be included in TEXT?

Good, I was going to ask it...  

> They don't fall into the syntactic definition of CTLs in 2616, but the  
> are semantically control characters, AFAIK.

...yes, unless the wannabe-Latin-1 is in fact windows-1252, UTF-8, or
what else.  Decree that C1 controls MUST NOT be used.  I hate this
approach, but if it's the only forward solution let's at least make
sure that wannabe-Latin-1 has a decent chance to really be Latin-1.

Maybe go as far as saying that 0x80 is no Euro in Latin-1, and that
while 0x85 might be a NEL elsewhere it is C1 and verboten.

> """
> comment = "(" *( ctext | quoted-pair | comment | encoded-word ) ")"
> """

Adapted from 2047 => okay.

> """
> field-content = <field content>
> ; the OCTETs making up the field-value,
> ; consisting of either *TEXT or combinations
> ; of token, separators, quoted-string and encoded-word,
> ; according to the syntax specified by the field.
> """

Maybe not good enough, for an unstructured *TEXT field
you need 2047 chapter 5 clause (1), no problem.

For a structured field you cannot permit encoded-word
as folks see fit, the structure must not be hidden by
encoding (parts of) it.

> """
> filename-parm = "filename" "=" quoted-string | encoded-word
> """

No 2231 folding here, might work.  Please note somewhere
that 2231 updated <encoded-word> adding optional language
tags.  Trying to forbid that could be bad.  

It is always tricky if A has a normative reference to B,
after B was updated or obsoleted by C, what does it mean
for A when working on A-bis ?  Eat our own dog food ?

> Also, I haven't addressed From (p2, 10.3). Anybody want
> to take a stab at that?

No Latin-1 in From.  For starters s/2822/2822upd/ minus
the obsolete syntax in chapter 4 of 2822upd (trust me,
you don't want anything obs-* ;-)

If you intend to evolve 2822upd From into some shrink
to fit 2047 MIME modulo 2231 forget it, only one person 
ever tried to combine 2882 + MIME, he is not here, and
he started with not less than four I-Ds full of hardcore
ABNF (nobody needs any prose or examples if the syntax
is "obvious")

 Frank
Received on Friday, 28 March 2008 03:06:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:50:37 GMT