Assuming the goal of namecoin and dot-bit is to put control over names in the hands of the users of the system rather than a centralized authority then we should avoid doing anything that precludes the use of alternate name resolution schemes (ie not DNS or namecoin) by the end user. This, I think, implies that we should avoid a proliferation of TLDs in order to avoid clashes (as happened with the altroot .biz domain). One problem with this
is that the dominant name resolution scheme (ICANN's DNS) has no such compunctions and indeed have financial incentives to cause such a proliferation. A way to resolve this would be to use a TLD that is illegal or at least highly deprecated under DNS rules but accepted by most software.
This could be achieved by using a TLD that ends in a dash which seems to work adequately. If trailing dashes were restricted to only the top (name service identifying) level then the presence of a label with a trailing dash could serve as an endianness marker.
Thus foo.b- and b-.foo (for example) would be the same name just written with a different endianness.
Not sure how this would work with IDN and its endianness rules.
Comments?
TLDs,other name resolution schemes and endianness
Re: TLDs,other name resolution schemes and endianness
This is an interesting idea. And more appealing considering this:
http://www.reuters.com/article/2011/06/ ... 9W20110617
If it is approved, someone could just buy the .bit TLD from ICANN...
EDIT:
In the spirit of testing, I added .- TLD support to ns.bit (DNS server at 78.47.86.43), so that people can test how and where it is supported. The result on the Linux resolver isn't satisfactory, but feel free to query the server yourself:
I also tried .b- and ._ , which give the same result.
With ICANN allowing arbitrary TLDs, it will be hard to avoid a conflict.
http://www.reuters.com/article/2011/06/ ... 9W20110617
If it is approved, someone could just buy the .bit TLD from ICANN...
EDIT:
In the spirit of testing, I added .- TLD support to ns.bit (DNS server at 78.47.86.43), so that people can test how and where it is supported. The result on the Linux resolver isn't satisfactory, but feel free to query the server yourself:
Code: Select all
$ dig example.-
;; ANSWER SECTION:
example.-. 300 IN A 208.116.53.211
$ ping example.-
ping: unknown host example.-
With ICANN allowing arbitrary TLDs, it will be hard to avoid a conflict.
bitcoin:13uSLCLqURqjJkfH6ny56h65oF6bfsVik1
namecoin:NEuYvWtQ8pZzFx7sAnfjeRVtzvieCWF5Ug
namecoin:NEuYvWtQ8pZzFx7sAnfjeRVtzvieCWF5Ug
Re: TLDs,other name resolution schemes and endianness
Got the same result here. My own testing involved putting a .b- name in my local /etc/hosts file where it worked both for ping and firefox. So it looks like both the protocol agnostic bits of the C library and the DNS protocol can cope with such names but the linkage between them somehow breaks it.