Re: [dnsext] afasterinternet.com trial and draft-vandergaast-edns-client-subnet-00

[SM <[e-mail]>]

Hi Wilmer,
At 14:10 30-08-2011, Wilmer van der Gaast wrote:
>I fully agree. By picking 0x50fa instead of a lower value (it looks
>like so far EDNS0 options are assigned sequentially) we certainly
>tried to avoid any problems of this kind. Our plan was to use this
>temporary option code just to gather data, and with that and a
>published experimental RFC, get an official number from IANA. Not at

You are running a trial over the Internet.  Someone out there will 
think that it is a good idea and write some code for it.  Such code 
generally lingers around for some time.

The above qualifies as an experiment and there was a time when it was 
documented through an Experimental RFC.  The code point was part of 
the book-keeping, i.e. the person is going to run the experiment no 
matter what and it is better to have a record of that code point 
being in use.  The number from IANA does not mean that the experiment 
or even proposed standard specification is a good idea.

Assuming that the assignment of code points is always sequential 
encourages "counter-measures" such as non-sequential assignments to 
make life difficult for people who make such assumptions.  There was 
a time when conventional wisdom was that the world would not run out 
of 32-bit numbers.  Some people looked at the way the assignments 
were made and decided that 1.1.1.1 could be used safely.

If it takes an Experimental RFC (published) to get that number, it is 
up to the IETF to see it wants to make it difficult for you to get 
it.  I doubt that people would refrain from using the technology if 
it makes an "Internet" faster.

Regards,
-sm