new nameserver - call for feature requests

[Ian Jackson]

I am in the process of designing and writing a nameserver, because I'm
not completely happy with any currently available Free nameserver
software.  I hope that the program, if and when I complete it, will
become the GNU nameserver.

The project is just starting; it's still at the malleable stage where
new feature requests and behaviours can be incorporated.  I'm
interested in suggestions from the community.  I have made certain
decisions already, and only some kind of suggestions will be useful,
so please first read this whole message !

* I've decided to have the following features, so there is no need to
  suggest them:
   - Full service resolver
   - Primary and secondary zone server
   - Forward (first or only)
   - Ability to get primary zone data from other sources (eg
     databases, programs) somehow, probably by compiling a module or
     some such into the nameserver
   - Provide, or don't, recursive service, depending on calling IP
   - IPv6 AAAA RRs

* I'll probably not include the following features in the first
  release, but may consider them later:
   - sending and receiving queries and answers via IPv6
   - attempting to hide some data from some queriers
   - DNSSEC (in whatever form it turns out to end up)

* There are two significant features I've decided not to include:
   - A6 RRs, bitstring labels, and DNAME (RFC2874)
      I have strong reservations about these; I think they should be
      abolished.  If you want to discuss them (for or against) in a
      general standards context, then the IETF ipng and/or dnsext
      working groups might be good places; I do follow much of their
      work and take an interest in these questions.
   - Two faced DNS inside the same server
      After considering this question in some depth, I have concluded
      that when two faced DNS is required, it is best implemented by
      running two instances of the nameserver software.
  Please don't argue with me about these decisions in private email;
  I have thought about these questions for some time, and I have
  reached fairly firm conclusions.  If you disagree, I can only offer
  my regrets that our views are not in alignment, and suggest that
  there is room for more than one Free name server project, and that
  there are other places where you can air your opinions.

* Any unobvious and not well known technical mistakes made in other
  nameserver software would be interesting to know about.

* Suggestions for implementation strategies are not very useful to me;
  I am doing the design myself.

* Advice on how to write secure software, or about other security
  questions, is not very useful to me.  I have much experience of (and
  a doctorate in) computer security.  If you have an interesting and
  novel observation or technique you should of course publish it in
  the appropriate place(s).

* If you know of existing Free nameserver software (for the relevant
  definition of Free see http://www.gnu.org/philosophy/free-sw.html)
  or projects to create such software please let me know.  I'm already
  aware of Dents.

* I am in complete agreement with RFC2826 `IAB Technical Comment on
  the Unique DNS Root', and do not wish to help with any `alternative
  root' schemes, or the like.  (This does not mean I support ICANN.)

* Offers of help are not needed at this time, thanks.

* Suggestions, queries or comments about licensing, or other
  organisational questions, are not helpful at this point.
  There is not yet any mailing list or website for this project.

* I'm afraid I can't promise a personal answer, since if I get many
  suggestions I may not have time to respond individually.  At this
  point I'm hoping that you'll provide input to help me write the most
  useful software, but I don't have time to get into many discussions
  about it.

* Please don't ask when it will be ready.  I don't know :-).  If and
  when it is suitable for early testing I will announce it in all the
  appropriate places.  Nagging me will at best have no effect, and may
  just put me off.  In any case, it will be ages, maybe years - don't
  hold your breath.

* Please send your comments to the Reply-To from this message,
  address@hidden  That email
  address is temporary, but will last until at least 13.5.2001.

Thanks in advance for your attention, and your help,
Ian.