On 18-Jul-2006, at 15:09, Douglas Otis wrote:
While wanting to agree, there is also an expectation in the DKIM WG that utilization of newer RR types will be problematic (especially with respect to MS), and a purported rationale for use of underscores and TXT.
Without wanting to drag the whole DKIM design discussion into this list (one list is more than enough, I think, for that) this thinking is thoroughly broken.
If the use of new custom-fit RRs causes implementation problems with specific platforms, then those implementations need to be fixed in order for the DNS to be extensible in the RRtype dimension. That's never going to happen if new RRtypes aren't routinely assigned and used for new applications.
The right thing for DKIM to do is to register a DKIM RRtype with appropriately-structured RDATA. DKIM's requirement for underscore names disappears as soon as there's a DKIM record which can be used, so DKIM *ought* not to be a consideration when thinking about an _names registry.
Which alternative would you expect to cause greater interoperability problems?
The requirement to parse free-form data from TXT RDATA is hardly a recipe for trivial interop. From the list of specification nits I saw in the first DKIM meeting relating to parsing the TXT RDATA, one might also conclude that it's not easy to specify, either.
Unless Microsoft has an enormous hidden marketshare in authority servers and ISP mail relays, I'm not sure why they're being singled out as problematic with respect to new RRtypes and DKIM, incidentally.
Joe . dnsop resources:_____________________________________________________ web user interface: http://darkwing.uoregon.edu/~llynch/dnsop.html mhonarc archive: http://darkwing.uoregon.edu/~llynch/dnsop/index.html
