[ietf-dkim] Issue 1576: dkim= no optimisation required

[Stephen Farrell]

Sorry, Bill, I meant to ask those who wanted change to
propose new text. In each case if the new text proposed
gets a sufficient bunch of +1's then we go from there,
but if not, we stay as-is.

Stephen.

Bill.Oxley at cox.com wrote:
> 
> -----Original Message-----
> From: ietf-dkim-bounces at mipassoc.org
> [mailto:ietf-dkim-bounces at mipassoc.org] On Behalf Of Stephen Farrell
> Sent: Tuesday, July 08, 2008 6:52 AM
> To: DKIM
> Subject: Re: [ietf-dkim] Issue 1576: Revise wildcard discussion
> 
> 
> So, trying to get to closure on this, rather than go
> 'round and 'round. There seem to be 3 things being
> discussed here, as listed below.
> 
> Even if you disagree and think there's a 4th thing - please
> hold your fire until we bottom out on these three.
> 
> We'd like to close out on these things by Friday (11th) so
> if you do care, you need to suggest text asap. If some
> suggested text seems to be getting a bunch of +1's then we
> won't be v. strict about that deadline, but if nothing is
> getting support then I think we'll declare victory on this
> one then.
> 
> Thing 1:
> 
> Charles Lindsey (and others) wrote (things like):
>> Having all genuine ADSP records  
>> start with some string such as "dkim=" will make such checks easier  
>> (though not foolproof because even a randomly created TXT record could
> 
>> start with "dkim-", though with low probability :-) ).
> 
> I think the counter John is making is that parsing the entire
> ADSP record is equally easy, so his argument as I read it is that
> the above is an optimization that isn't worthwhile.
> 
> Is it really significantly better/easier to check starts-with("dkim=")
> rather than matches(the-ABNF-from-4.2.1")? The ABNF from 4.2.1. is:
>              adsp-dkim-tag = %x64.6b.69.6d *FWS "=" *FWS
>                              ("unknown" / "all" / "discardable")
> 
> (If responding, please ignore the [FWS] vs. *FWS issue.)
> 
> If you think the optimization is worthwhile, please post
> text (and not discussion) and we'll see if that gets +1's
> or -1's or gets ignored (at this stage I think the latter
> two are the same). If you do post text for that please put
> "dkim=" in the subject as well as 1576.
> 
> Thing 2:
> 
> Currently ssp-04 says that "ADSP records use the "tag=value"
> syntax..." and also says that only one ADSP record is to be
> published.
> 
> So if that means that records returned because of wildcards that
> don't match eactly one instance of the ABNF are ignored then we seem
> to be ok. If however, the current text isn't explicit enough about
> how to handle records that don't parse correctly, or multiple records
> then maybe there's more to say.
> 
> If you think there is, then please post text (and not discussion)
> and we'll see if that gets +1's or -1's or gets ignored (at this
> stage I think the latter two are the same).  If you do post text
> for that please put "handling" in the subject as well as 1576.
> 
> Thing 3:
> 
> The last thing that seems to be involved in this thread is a
> concern that the current description of the dangers of wildcards
> is not sufficiently well explained. If you think that that is
> the case then please post text (and not discussion)
> and we'll see if that gets +1's or -1's or gets ignored (at this
> stage I think the latter two are the same). If you do post text
> for that please put "caveats" in the subject as well as 1576.
> 
> Thanks,
> S.
> 
> 
> _______________________________________________
> NOTE WELL: This list operates according to 
> http://mipassoc.org/dkim/ietf-list-rules.html
>