[LEAPSECS] the DNS idea is old
seaman at noao.edu
Thu Apr 9 09:48:18 EDT 2015
Thanks for the reminder!
This was announced on the original leapsecs list on 24 Apr 2003:
(So one or the other is likely misdated.) Useful discussion followed:
(The file numbering is reversed; a good way to scan Steve's resurrection of the original list is by date: http://www.ucolick.org/~sla/navyls/date.html)
Poul-Henning's idea to use class E is an answer to Clive's objections (at the time, and in the past few months). There are good reasons to prefer class E to TXT records, though the latter is better for conveying lists.
A checksum is indeed significant. But zefram's formulation using SHA is more general purpose, e.g., you can cut it or the string at any size without retuning against a different polynomial. For example, does the same polynomial apply to IPv6 as IPv4? Or TXT records might want a full cryptographic hash. But the best choice for a hash will be whichever one first has interoperating reference implementations in Perl, Python, C/C++, ... ;-)
> On Apr 9, 2015, at 12:25 AM, Poul-Henning Kamp <phk at phk.freebsd.dk> wrote:
> Just setting the record straight: I was not the first person to think
> of using IPv4 addresses via DNS as a leap-second communication mechanism:
> Dated 2002-04-24.
> However, I do think my inclusion of a CRC-8 sum is a significant improvement.
> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
> phk at FreeBSD.ORG | TCP/IP since RFC 956
> FreeBSD committer | BSD since 4.3-tahoe
> Never attribute to malice what can adequately be explained by incompetence.
> LEAPSECS mailing list
> LEAPSECS at leapsecond.com
More information about the LEAPSECS