Re: Non destructive folding/unfolding (ie, no insertion ofwhitespace in unfolding)

At 01:35 01/11/2007, Mikel Lindsaar wrote:

>Hello list,
>
>This is my first post, so I apologise if this has already been
>covered.  I searched through the full text archive file of ietf-822
>(after going through the full itef-smtp archive file) and could not
>find a concrete answer to my question.
>
>I am improving/fixing bugs in a Email library for Ruby - I didn't
>originally write the library, so I am being very careful in any
>changes I make.
>
>One situation I have is where a X-header field and Reply-To field has
>a single piece of text that is longer than the 78 octet limit, but has
>no white space break.
>
>For example:
>
>Reply-To: 
><1234-1234-1234-1234-1234-1234-1234-1234-1234@me-some-host-over-here.org>
>
>The library currently folds this line as:
>
>Reply-To:CRLF
>LWSP<1234-1234-1234-1234...blah-with-no-further-fold
>
>Now, I have two questions.
>
>1) I believe that putting the CRLF immediately after the : in the
>header field name is in violation of 2822 specifically that the header
>line should contain the name, a colon, the field and terminated by a
>CRLF.  Would this be a correct interpretation?

As I understood it this was OK. For a general unstructured header 
field, it begins with optional FWS, so can be folded there. The 
'Reply-To' is a bit different because it's structured, but I thought 
that FWS could be added at will around the address fields in a 
reply-to field. (I can't find the description of this at the moment 
in RFC 2822, but I'm pretty sure it's OK as that's what section 2.2.2 
seems to say

>2) Irregardless of (1), what is the correct handling of a piece of
>text in the field that is longer than 78 characters, but contains no
>whitespace to fold?  ie, in unfolding, no whitespace can be inserted
>back into the original text which would be destructive to the original
>text.

Just leave it longer than 78

78 is a 'soft' limit, 998 is the hard limit.

Section 2.1.1 -
Each line of characters MUST be no more than
    998 characters, and SHOULD be no more than 78 characters, excluding
    the CRLF

>I saw one mention of doing this:
>
>Reply-To: <1234-1234-1234-1234-1234-12\CRLF
>LWSP34-1234-1234-1234

NO - this will break things. You can only fold at FWS, you can't 
randomly insert FWS into a structured header field...