Stephen Farrell wrote:
> Issue descritption: https://rt.psg.com/Ticket/Display.html?id=1543
>
> ssp-04 still includes the FWS, I didn't see a clear consensus on the
> list either way (or else I missed some messages;-)
>
> If we really need a strawpoll/consensus check on this please say so
> and I'll start one.
>
> Otherwise, I'll assume its ok to stick with FWS and ask Eliot to
> close this one on July 11.
>   
To clarify what happened, I had changed [FWS] to *WSP in the ssp-02 
ABNF, in response to this issue.  It stayed that way in -03.  In 
response to Eric's comment below (forwarded with permission), John 
changed it to *FWS in -04.  I think we're agreed that [FWS] is more 
correct than *FWS, but there's probably still some difference of opinion 
on [FWS] vs. *WSP.  Here (I think) are the arguments on both sides:

- Favoring *WSP:  It doesn't make sense to require a leading whitespace 
character following CR/LF in a DNS record, since the leading whitespace 
applies to identifying continuation lines in RFC 2822 header fields.

- Favoring [FWS]:  This is the ABNF used in RFC 4871, and it doesn't 
make sense to introduce a different parsing requirement for ADSP.

Thoughts?

-Jim

-------- Original Message --------
Subject:        Re: [SA]SP ABNF for white space
Date:   Sun, 24 Feb 2008 11:38:45 -0800
From:   Eric Allman <[EMAIL PROTECTED]>
To:     Jim Fenton <[EMAIL PROTECTED]>
CC:     John Levine <[EMAIL PROTECTED]>
References:     <[EMAIL PROTECTED]> 
<[EMAIL PROTECTED]>



Jim,

I'm willing to discuss the point of what we should have done in 4871, 
but that's not the topic right now.  The fact is that 4871 already 
permits folding in TXT records and ssp-02 does not.  Implementations 
really should be able to use the same parser for these two cases, 
since they are identical.  Making them different is begging for 
implementation inconsistency.

eric


--On February 24, 2008 10:16:16 AM -0800 Jim Fenton 
<[EMAIL PROTECTED]> wrote:

> Eric,
>
> Yes, I changed this in response to an issue raised by Frank
> Ellermann.  I didn't see how the mail-header convention of
> continuing lines by putting whitespace at the beginning of the
> continuation line made sense in a DNS TXT record.  There was some
> discussion about leaving out whitespace in the ABNF altogether, but
> I thought it helped readability to put in optional whitespace,
> *WSP, in order to make sure implementations reading the record
> could handle whitespace because somebody's going to put it in
> anyway.
>
> Sorry if this is a surprise; this happened during your downtime.  I
> don't think there are any other surprises like this though.
>
> -Jim
>
> Eric Allman wrote:
>> Jim,
>>
>> I just noticed that the ABNF changed between ssp-01 and ssp-02 to
>> use  *WSP instead of [FWS].  These are semantically different, and
>> 4871  uses [FWS].  In particular, the definition of tag-list in
>> 4871 uses  [FWS], which means that there is a conflict in ssp-02,
>> which says that  the tag-list is the same syntax as in 4871 but
>> then defines the  individual tags using a different syntax.
>>
>> Sorry if this isn't clear.  I did want to check with you before I
>> posted something to the list in case I'm missing something obvious.
>>
>> eric
>>
>




_______________________________________________
NOTE WELL: This list operates according to 
http://mipassoc.org/dkim/ietf-list-rules.html

Reply via email to