bortzmeyer

import unicodedata

def escape(ustr):
    result = "" 
    for letter in ustr:
        if unicodedata.category(letter) in ['Ll', 'Lm', 'Lo', 'Lt', 'Lu', 'Nd', 'Nl', 'No']:
            result += str(letter)
        elif letter == '&':
            result += '-and-'
        elif letter == '|':

bortzmeyer

def escape(ustr):
    """ Escape an Unicode string to keep only characters which are safe for a filename. """
    result = []
    for letter in ustr:
        # Keep only letters and numbers                                              
        if unicodedata.category(letter) in ['Ll', 'Lm', 'Lo', 'Lt', 'Lu', 'Nd', 'Nl', 'No']:
            result.append(str(letter))
        elif letter == '&':
            result.append('-and-')
        elif letter == '|':

bortzmeyer

% traceroute6 www.cogentco.com 
traceroute to cogentco.com (2001:550:1::cc01) from 2001:470:1f10:3aa::2, 30 hops max, 24 byte packets
 1  bortzmeyer-1.tunnel.tserv9.chi1.ipv6.he.net (2001:470:1f10:3aa::1)  10 ms  19.999 ms  9.999 ms
 2  gige-g3-4.core1.chi1.he.net (2001:470:0:6e::1)  9.999 ms  10 ms  9.999 ms
 3  10gigabitethernet3-2.core1.den1.he.net (2001:470:0:1af::2)  29.999 ms  29.998 ms  49.997 ms
 4  10gigabitethernet4-4.core1.sjc2.he.net (2001:470:0:1b4::1)  49.997 ms  69.996 ms  59.997 ms
 5  10gigabitethernet3-2.core1.pao1.he.net (2001:470:0:32::2)  59.996 ms  69.996 ms  49.997 ms
 6  * * *
 7  * * *

bortzmeyer

Terminology : why "registrar"? From the description, it is a registry.

Requirements: apparently, no explicit list of requirments, no specific
list of flaws identified in DNS or in other projects.
 
Why only one registry? This will make it the focus of many attacks,
technical and political. Anyway, I'm ultra-skeptic about this idea of
an ideal registry which can be trusted. Too much concentration of
power. That's the biggest problem with INS.

bortzmeyer

LogFormat "[%h]:%{remote}p %l %u %t \"%r\" %>s %b \"%{Referer}i\" \"%{User-Agent}i\" %v" combinedv6
LogFormat "%h:%{remote}p %l %u %t \"%r\" %>s %b \"%{Referer}i\" \"%{User-Agent}i\" %v" combinedv4

SetEnvIf Remote_Addr : remote-v6
CustomLog /var/log/apache2/access.log combinedv6 env=remote-v6
CustomLog /var/log/apache2/access.log combinedv4 env=!remote-v6

bortzmeyer

status.net ne permet pas de vraie fédération car il n'existe pas de mécanisme d'identificateur 
unique. Si deux personnes lancent une instance de status.net et se nomment "jean", ils ne pourront 
pas connecter leurs instances car l'identificateur "jean" sera ambigu. Et s'ils se connectent 
ensuite à identi.ca et que celui-ci a aussi un "jean", même problème.

De ce point de vue, status.net est en retard sur XMPP (où l'identificateur est nom@domaine et est 
donc unique dans tout l'Internet) et même par rapport au courrier électronique qui a ce système 
d'identificateurs fédérés depuis quarante ans  !

bortzmeyer

> est-ce utile de mettre à la fois "ip4:ip.du.server.master", "a" et "a:mail.example.tld" si tous 
> pointent sur la même machine ?

Cela peut servir en cas de changement de configuration. Des tas de gens ont mis des beaux enregistrements SPF, les ont oublié, et ont tout cassé en changeant une adresse IP. Toujours penser au futur !

> y a t-il une nécessité de mettre "mx:antispam.tld" si cette machine ne fait que recevoir 
> des messages (je ne pense pas).

Non, je ne vois en effet pas de raison de le mettre.

bortzmeyer

#include <stdbool.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <netdb.h>
#include <arpa/inet.h>
#include <errno.h>
#include <netinet/in.h>
#include <netinet/ip.h>
#include <netinet/ip6.h>

bortzmeyer

DIVX0929<?xml version="1.0" encoding="UTF-8"?>
<cduparameterGenerator:CallParameters xmlns:cduparameterGenerator="com.divx.util.parame
terGenerator." skip="false"><cduparameterGenerator:Parameter name="nullParam000" paramN
ame="fileType" value="Ticket" /><cduparameterGenerator:Parameter name="nullParam001" pa
ramName="fileTypeVersion" value="2.1" /><cduparameterGenerator:Parameter name="nullPara
m002" paramName="ticket" value="PaCWvsfAihIUUMQAvTnkOz7_MlMerylX-1311710199895" /><cdup
arameterGenerator:Parameter name="nullParam003" paramName="title" value="Alice au Pays 
des Merve illes" /><cduparameterGenerator:Parameter name="nullParam004" paramName="cont
entId" value="1252451" /><cduparameterGenerator:Parameter name="nullParam005" paramName
="dpcPlaybackAllowedInd" value="true" /><cduparameterGenerator:Parameter name="nullPara

bortzmeyer

After reading the two excellent papers mentioned as reference, and after discussing the matter with several persons, I tend to think that the fundamental problem is the existence of several user communities, each with different (but perfectly legitimate) requirments. Because these requirments are different, there is zero chance to find a way to satisy them all with one time scale. The only solution is therefore to have several scales and to let each community choose the one which fits its requirments.

Of course, there are already several time scales. I feel that most of the needs of people who want the end of leap seconds would be satisfied by TAI (a very regular time scale, without "steps" and without link with the solar time). If, for one reason ot the other, TAI is not perfect for them, and there is no existing time scale suitable, it may be interesting to develop a new time scale (I'm not convinced it will be necessary: many proposals, such as the one for the "new UTC", are YATSCOT - "Yet Another Time Sc