[dns-operations] creeping poorness of judgement

John Levine johnl at taugh.com
Sun Mar 15 00:47:45 UTC 2020


In article <c8c54222215ad185d59dd74abd856230 at ausics.net> you write:
>So if the issue is the TXT entry cant hold what you need it to hold,
>then it needs to be more efficient. See how google does it - it works. 

Sorry, I don't understand your point, unless you're suggesting people
use the way that Google breaks up its SPF into multiple records with
different names with SPF includes to combine them.  I guess that's OK
but I don't see much merit in using four records with four names to
represent info that would just as well fit into one record with one
name.  I presume Google can provision TXT records with multiple
strings.

>Lastly, TXT and SPF - blame debians Scott Kitterman, he was the one who
>so furiously argued agaisnt SPF having its own RR, he is the one
>responsible for the massive push that saw it junked. 

That's a rather egregious rewrite of history.  When RFC 4408 was
written, SPF was already widely implemented using TXT records.  The
DNS mafia of the era held the document hostage until the authors
agreed to add a type 99 SPF record.  In practice, nobody ever
published type 99 records, other than a few of us who added hacks to
our DNS crudware to mirror TXT records that started with v=spf1.

Six years later, RFC 6686, which Scott did not write, surveyed over a
million domains with MX records and found that rounded to the nearest
percent, nobody published type 99 records, so that was that.



More information about the dns-operations mailing list