spf-discuss
[Top] [All Lists]

Re: draft-schlitt-spf-00pre4 now available

2004-11-05 11:27:49
wayne wrote:

*UGH*  This is getting overly complex.
[...]
now four different things?  IMO this is unnecessary.  A
global counter makes more sense.
[...]
I agree that the wording in the spec is somewhat messy, but
in practice, publishers just need to count the number of
mechanisms and make sure they have less than 10.

Then just say so, and get rid of these implementation details.

You could add Caveats to the chapters with MX and PTR for SPF
publishers, and allow a general "panic - resources exhausted"
error for odd cases where "10 mechanisms" isn't good enough.

It should be 10 mechanisms minus (ip4/ip6/all) plus redirect=

I don't think SPF records like "v=spf1 a (#*LSLK$" should
sometimes return a pass and sometimes return a syntax error.

Yes, that's ugly in draft-lentczner-spf-00.

I don't like the fuzzyness of whether "v=spf1 -all
accredit=_spf.%(d)" means that the syntax error in the macro
variable is "encountered" because you you have to look for
the "exp=" modifier.

Is that a syntax error ?  An implementation knowing shit about
accredit= should be free to ignore "unknown modifiers".  For
exp= it would be wrong.  If I'd implement SPF I'd ignore any
exp= because I hate the idea, no matter what one of the spec.s
says about the exp= syntax.

I'm not going to try and fight Meng and MarkL in the IETF
over which draft should be used.

Somebody else will unless Mark reappears within the next days.
Getting it right is more important than personal vanities.

                    Bye, Frank