[OpenAFS] Re: is there any good reason to use capialized names for new realms?

Russ Allbery rra@stanford.edu
Thu, 26 Jan 2006 21:11:50 -0800

Adam Megacz <megacz@cs.berkeley.edu> writes:
> Russ Allbery <rra@stanford.edu> writes:

>> And third, the assumption isn't in the applications; it's in libkrb5.
>> Upcasing the domain is the final fallback algorithm used to determine
>> the realm for a host if DNS lookups fail (or are turned off, which is
>> the default as mentioned by others) and there's no realm mapping in the
>> krb5.conf.

> I see.  Where can I find the official standard procedure by which this
> mapping is performed by all compliant implementations?

Hm, I'm not sure that anything like that *could* exist.  The RFCs are
concerned with the on-the-wire protocol; locally configured mappings would
be outside their scope.  MIT Kerberos and Heimdal are different
implementations that could well use different logic.  Shishi is yet
another implementation with its own logic.  Microsoft might do something

I don't think this is the kind of area where you are going to get that
sort of document.

Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>