2007-12-01 La Naissance Des Pieuvres
During the European Movie Festival of Virton, we saw various good films but I was really impressed by the film La Naissance Des Pieuvres made by Céline Sciamma. I choose the film based only on its title and that was a clever choice. Often people are going to the cinema (or the movie theather in US) and expecting action, impressive visual effect without taking their breath between each action. With La Naissance Des Pieuvres, this is completely different and that's wonderful. The film talks about the entrance to the "adult" world by three teenagers. This takes place around a swimming pool and a synchronized swimming club. Nothing more… but that's the great part, with near nothing, Céline Sciamme (and their actors) expressed a lot. I was really touched by the film, the actors, the situation and the music. A good example of what I'm expecting from a film : thinking about ourself. The film (and its action) was still in my mind after some days. That's usually a very good point for a film at least in my eyes. I also discovered that the electronic music was made by Para One aka Jean-Baptiste de Laubier and the music is also a strong point in La Naissance Des Pieuvres. I purchased the original soundtrack and discovered the music flow created for the film. Very interesting, the overall music theme is repeating until you reach the track named 'Finale'. A great moment…
Tags : film movie electronic_music electronicmusic paraone celinesciamma life electro
2007-12-09 Innovation at Work
Via Scott Berkun, I discovered the paper of Allan L. Scherr about Managing for Breakthroughs in Productivity (PDF). I'm always very sceptical when it's about "managing innovation" as I tend to consider this is very difficult to control something often unpredictable. Allan L. Scherr in his paper rejects that idea of "difficulty to control innovation" (first described in Breakthroughs! written by John M. Ketteringham and P. Ranganath Nayak) and explain in 20 pages how to manage productivity (and innovation ?). The paper is really well written and concise to describe the concept. I really loved this part of the conclusion :
The culture of breakthrough projects can be spread and productivity improvements have been seen in adjacent business-as-usual groups. The key to spreading this culture is upper management support for taking the risks that are inherent in such endeavors. If management cannot tolerate risk and the occasional failures that occur, it does not take long for the culture to return to conventional, business-as-usual ways.
When talking about innovation in companies, this is a critical part : the ability to take risks inside the company. In my eyes, this is valid on both approaches : managed and unmanaged innovation. If your boss (or you) are able to take risks, that's maybe the beginning for some innovation…
Tags : innovation startup work productivity
2007-12-10 DNS Suffix Lists Considered Harmful
A lot of name resolver have a kind of suffix lists (or search list) to lookup when trying to resolve a non-FQDN hostname. This use is quite common in internal (e.g. in large company) networks to ease the typing of people. Like that people can type intranet in their favourite browser instead of typing the FQDN like intranet.mycompany.internal. In theory, this looks nice and lazy users are quite happy. In practise, this is a nightmare… One the common example, company have a global suffix configured on all internal desktop computers like :
.dyn.mycompany.internal .mycompany.internal
Just imagine, a simple misconfigured dynamic name client named "intranet". In such order, the intranet.dyn.mycompany.internal will be resolved before intranet.mycompany.internal. The easy solution will be to change the order of the search list to avoid the described scenario. Yes this could solve a part of the issue as long as the user is not setting up is own suffix list. The suffix can be also received dynamically if the DHCP server is supporting the RFC3397 as an extension (especially look at section 4. Security Consideration) . That adds some complexity in the potential scenario of the name resolution and that's not good on a security perspective. Just have a look at the flowchart made in the name resolution documentation from Microsoft. You have a variety of case as long as the name resolution software is behaving as expected. The situation is somehow the same with the search list in the Unix-like world. Do we still need suffix search list for name resolution ? does it help or is this adding too much potential issues (from security to simple network debugging) ? If someone is asking me about it, just remove all search list (static or dynamic) and inform users to always use FQDN.
Tags : name_resolution dns security fqdn system_administration
2007-12-15 Publish Or Perish
Reading the latest viewpoint of David Lorge Parnas in the Communications of the ACM (November 2007/Vol. 50, No. 11), he is annoyed by the increasing number of publication in computer science with less scientific value. As the measurement in research is often mainly based by the number of papers published instead of the quality and real contribution in the publication. As an example PhD students often prefer to publish recurring paper on a same topic than having a single high-quality paper. This is mainly due to an evaluation done by only checking the number of papers than really digging into each paper made by the student. This is an old issue in scientific publication or in any single metric to measure research, productivity or innovation (as I already discussed in Innovation Metric and using a single metric like the patent system for measuring innovation).
I tend to agree with Mr. Parnas that counting papers just slows the rate of scientific progress. But the source of the issue is not only the "the Numbers Game" but it's the overall (closed) peer-reviewing process. As the reviewing is done in "island mode", there are not shared not only the paper but also the reviews of it. An open archive process permits to "pre-publish" in order to peer review publication in advance. I really like the idea to have a kind of continuous on a publication. That could open the doors of more exchange between people in the computer science area. For example , the arXiv.org open archive project is already providing a nice interface and permits to trace the submission history. That's maybe the beginning of something new. That could also improve the current situation in conference where people come to make a presentation only… and that's it. Without any exchange or discussion between the participants. I'm not negative just here to improve a little bit the current situation.
2007-12-22 IPv6 Village Experiment
At home, we have a small guest wired/wireless network for the guest addicted (nearly everyone is addicted including myself) to plug their laptop into the global Internet. Willing to validate the reality of IPv6 with Free Software, I move that network to an IPv6-only network. I took some notes about the installation and the various tests. Of course, the network is only IPv6 but we didn't want to put them on an island without accessibility to the majority of services available in IPv4 only.
First of all, you need a good connectivity to the IPv6 Internet. Bad luck in Belgium, there are no providers supporting natively IPv6 until now. So, we used the sixxs IPv6 tunnel broker service and use the a part of the /48 subnet allocated to me by sixxs. I split the subnets and use a simple /58 (still big enough ;-) subnet for the guest network. I have a gateway machine running a standard Ubuntu 7.10 and running the sixxs tunnel client (until Belgacom Skynet is providing IPv6 to their customers). The sixxs tunnel client works great behind a NAT of the IPv4 traffic. An Ethernet interface is allocated for that guestnet and doing the routing with the ipv6 tunnel interface. That's great with that setup my guest can access all the IPv6 internet and everyone can reach them (real public IP addresses for the client). I have also setup on the gateway machine an ISC bind 9 nameserver answering queries in IPv4 or IPv6 (that's the standard nameserver in Ubuntu without custom configuration). Great… but in such scenario, they have only access to services available in IPv6.
So we are going back to the initial chicken-egg problem for the IPv6-to-IPv4 transition… I didn't want to have a dual-stack guest network and really want to provide a full-blown connectivity in IPv6. So the compromise is to provide an HTTP proxy supporting IPv6 seamlessly. My first thought was to go with Squid but they only introduced IPv6 in their HEAD branch the 16th Dec. 2007. I compiled it (--enable-ipv6) but it works but I had some issue with the behavior of "tcp_outgoing_address" feature to reach IPv6 accessible without using the IPv4 connectivity. Then I discovered an alternative proxy, written by Juliusz Chroboczek and called Polipo (also available in any recent GNU/Linux distribution) with a good and coherent IPv6/IPv4 support. With that my IPv6 guests are now able to talk natively to the IPv6 Internet and use the proxy to reach the IPv4 Internet (mainly HTTP,FTP or TLS/SSL).
Beside that IPv6 development is quite old, the real roll out of IPv6 in common (free and non-free) operating system is now a reality. That offers the possibility to have an IPv6 connectivity in place quite quickly without changing the underlying operating system. There are still some glitches in some top application managing network connectivity without knowing the existence of the IPv6 connectivity but it works. The IPv4 Internet can be easily reached using a proxy. The main advantage of having a native IPv6 connectivity is to have real routable addressing and starts to provide services to the IPv6 internet (e.g. just imagine distributed network sensors).
After my very little success for the guest network, why not providing the IPv6 service to the whole village ? I just need to extend the reachability of my wireless network (not impossible looking at the size of the village). Digging a little bit of an allocation plan using DHCPv6 (better than the a simple "autoconfiguration" lacking network service information) and doing some promotion…
Tags: ipv6 networking freesoftware village ipv4
2007-12-27 RSS Everything XMAS
After the past RSS Everything blog entry made in February about RSS and the catrss from Jean-Etienne Poirrier, I come today with two new experiments for my own need (I still hope one day to generalize the script to make them more UNIX-like filtering friendly).
I wanted to merge my multiple activities (often represented in RSS) in one representation to give an overview to be accessible on my personal home page. I made a simple script rssmerge.py to do the job. A sample output result is available.
python2.5 rssmerge.py --maxitem 200 --output phtml "http://www.foo.be/cgi-bin/wiki.pl?action=journal&tile=AdulauMessyDesk" "http://api.flickr.com/services/feeds/photos_public.gne?id=31797858@N00&lang=en-us&format=atom" "http://a.6f2.net/cgi-bin/gitweb.cgi?p=adulau/.git;a=rss" "http://www.librarything.com/rss/reviews/adulau"
The other script (rsscluster.py) is also a very early experiment to cluster by time multiple items from an RSS feed into another RSS feed. It's quite common to have an RSS feed containing a lot of items for small events (like bookmarks) and you want to cluster them in one item for a 2 days period.
python2.5 rsscluster.py --interval 2 --maxitem 20 "http://del.icio.us/rss/adulau">adulau-2days.xml
The two scripts are accessible in my messy git repo. No major innovation but still small step to better (or more?) use RSS (at least for me). I also hope that will trigger more discussions or other ideas around RSS (or Atom).
2007-12-30 My Favorite RFCs
It's quite common to see a list of favorite at the end of year. Everyone has favorite… I'll make a list of my favorite RFCs. This is not really a favorite list as it's often boring to read (and to really understand/interpret… the most difficult part) RFCs but you are obliged if you want to implement or review software that should follow them. I'll make a quick list of the RFCs that I read the most the past two years (this is not always reflecting the quality of the RFCs but more the usefulness of them to my past works) :
- RFC 2616 - The Hypertext Transfer Protocol is everywhere but not always strictly followed by a lot of software. My favorite section is 14.36 with the misspelled "Referer". A lot of caching proxies have a different understanding of the RFC and especially with the "SHOULD"… HTTP is more and more used, just have a look at REST strongly relying on HTTP.
- RFC 2440 (and by extension RFC 4880) - The OpenPGP Message Format - not easy to master but with a companion implementation like GnuPG, this helps a lot.
- RFC 2196 - Site Security Handbook - A great and simple reference for implementing security with a pragmatic approach without the hassle of ISO 27001 or ISO 27002.
- RFC 2396 - Uniform Resource Identifiers (URI) - somehow related to RFC 2616 but a must to use if you write crawler that need to parse a lot of URI. You could be very surprised by the wide definition for URI in practice.
I have some more but I won't add them in the list because I had already a bunch of nightmare by just trying to figure out the difference between the specification and its implementation.
On the fun side (you know the 1st April RFCs), my favorite is RFC 2795 (The Infinite Monkey Protocol Suite (IMPS)) and especially the PAN (Protocol for Assessment of Novelty) section with a nice (and sometime useful) CRITIC reject code :
8.3. Table of CRITIC Reject Codes CODE DESCRIPTION ------------------------------------------------------------------- | 0 | <Encrypted response following; see below> ------------------------------------------------------------------- | 1 | "You're reinventing the wheel." ------------------------------------------------------------------- | 2 | "This will never, ever sell." ------------------------------------------------------------------- | 3 | "Huh? I don't understand this at all." ------------------------------------------------------------------- | 4 | "You forgot one little obscure reference from twenty years | | ago that renders your whole idea null and void." ------------------------------------------------------------------- | 5 | "Due to the number of submissions, we could not accept every | | transcript." ------------------------------------------------------------------- | 6 | "There aren't enough charts and graphs. Where is the color?" ------------------------------------------------------------------- | 7 | "I'm cranky and decided to take it out on you." ------------------------------------------------------------------- | 8 | "This is not in within the scope of what we are looking for." ------------------------------------------------------------------- | 9 | "This is too derivative." ------------------------------------------------------------------- |10 | "Your submission was received after the deadline. Try again | | next year." -------------------------------------------------------------------
If you are familiar with the WG at IETF with any technical meeting. It's quite common to see meeting interaction using a variety of CRITIC reject code. So most innovative people are often using a variety of them or sometime create new ones. I'm just wondering where is the IANA consideration for this table ? ;-) just if we need to update it…
![[FSF
Associate Member]](http://www.foo.be/fsfbutton-adulau.png){kind=small}
