[prev in list] [next in list] [prev in thread] [next in thread] 

List:       namedroppers
Subject:    Re: work items
From:       Masataka Ohta <mohta () necom830 ! hpcl ! titech ! ac ! jp>
Date:       1997-09-04 18:17:32
[Download RAW message or body]

Randy;

> >   It is known that interoperability problems occur in the operational DNS
> >   infrastructure when new RR types carry domain names in the RDATA section
> >   and name compression is used.  Not using name compression does not
> >   restrict the functionality that can be achieved by defining new RR
> >   types.
> 
> Although this seems to be the most workable suggestion to date, two aspects
> cause me to hesitate.
> 
> It puts a decision process on the IANA.  I will be interested in the IANA's
> perception of this.
> 
> The last sentence is not entirely true because of <bleeping> UDP MTU.  And,
> at the same time we are discussing this approach, we are dealing with the
> how-much-fits-in-a-packet discussion elsewhere.  

How about the fllowing scheme?

	RR type

	X~X+Y		RDARA section contains no domain name
	X+Y~X+2*Y	RDATA section contains one domain name at
			tbe beginning of the section
	X+2*Y~X+3*Y	RDATA section contains two domain names at
			tbe beginning of the section
	X+3*Y~X+4*Y	RDATA section contains three domain names at
			tbe beginning of the section
	X+4*Y~X+5*Y	RDATA contains variable number of domain
			names. The first two byte is a big-endean
			integer of the number of domain names, immediately
			followed by domain names.

In all cases, other data may be contained in RDATA after domain names.

Y could be as large as 8K.

							Masataka Ohta

[prev in list] [next in list] [prev in thread] [next in thread]