Izumi Blog: Ralf - Random Thoughts
|
Index: Home | What Is Izumi | Misc Links | Random Thoughts | Too Much To Read | The Rant Vault | Quotes
Dev: Projects | Ideas For Dev | Nerdkill | Rig | Hint
This place contains misc remarks and notes.
They are not quite rants though :-).
This blog page uses reverse-date ordering.
Site License And Disclaimer as well as contact information are available here.
$Id: Random Thoughts.izu,v 1.22 2005/11/05 21:55:57 ralf Exp $
«» 2004/05/01 «»
In Digital Imprimatur
and Speak Freely end of life announcement,
John Walker argues that broadband and NATs/firewalls are happilly
killing the ability of people to create servers of their own.
I agree with that. Yet, there are solutions against or around it.
Defeating without trying to make available a solution doesn't seem right here (after all
he was actively advocating Speak Freely).
I see 3 solutions to the problem. Let's summarize the problem first.
NATs or firewalls are not the problem. They won't go away,
even if we ever get IPv6 deployed massively. Today I do have a DSL with an fixed IP. It costs
a lot ot get that, a lot more than a regular user with a dynamic IP. I could get more IPs
from my provider for more money, but I don't want to. Having only one fixed IP costs less.
Period. Even if IPv6 was here to solve any kind of IP shortage, I would still expect my
provider to charge more for a netmask rather than a single IP and still more for a single
fixed IP than for a dynamic IP.
Another part of the problem is that I do manage my own DSL server, so I do control my firewall
and router. I can tweak it the way I want. And I want all my boxes behind to be on a NAT,
mainly because having them on a separate subnet means they are not directly accessible
from the outside, and given the crap I see hit my firewall I prefer it that way.
So NATs and firewalls are not going away. I do actually argue that it's better that way.
I can manage my own little network the very way I want it, without having to involve my ISP.
The only problem is that Mr. Clueless User can't run a server on his PC behind his broadband
modem if the ISP decided this modem box should perform NAT/firewalling stuff without the
user having a choice. Some broadband modems can do this NAT/firewalling stuff, but they don't
all do.
I see 4 possibiliies here:
- Say "no thanks you very much" to the ISP and choose another one that provides a modem that does not lock the user. More on this below.
- Force manufacturers to implement forwarding protocols on the modem box, for example RSIP (RFC 3102, RFC 3103 and RSIP for Linux).
- Use a simple modem and a Linux box that you can control to handle NAT and firewalling.
- Find a technical way to hack around connection limits.
I do agree with Digital Imprimatur
that this situation exists mostly because users don't care. Most users will probably go
for a cheaper service with dynamic IP rather that pay a premium for a fixed IP. Either
they do not understand what they loose or they simple don't loose anything. Most normal
users I know do not care about running a server on their computer at home -- well that
recently changed since a growing percentage of users were running p2p filesharing servers,
but I'm not even sure here they understand the difference between server and client in a
p2p filesharing world; heck most will be happy to simply download and never give back, which
is why some protocols make the download bandwidth a factor depending on the upload bandwith.
Now add to thist that most normal users will turn their computers off when not using them
(at at least expect them to go to sleep and not be active).
I really think those who consciously want to share are a minority.
I mentionned the Linux box stuff above. If we really wanted to make it possible for normal
people (not geeks that spend their nights RTFMing), the best thing we could do is create
a Linux distro that can work as a broadband server with easy setup of NAT, firewalling
and forward connections. I mean easy to the point that it would work out of the box for a
vast majority of users, with a smart and simple interface for newbies (and a way for advanced
users to get more advanced options).
Heck, I do maintain my Debian distro and this is plain painful. Sure it's a great OS, open
source movement blah blah, but it's bare to the metal. I understand most of the stuff that
runs on it yet sometimes I need some well advised external help to solve difficult networking
issues. And come on, installing this from scratch sucks. It's plain hard for me, yet I am
willing to read a lot of docs and willing to spend more than my share on free time on it.
So really, having a Linux distro purposedly targetted as being a simple but efficient
broadband router/NAT/firewall with an easy to use interface would really help the cause.
Then it would be simple to tell people, look get rid of the evil ISP and the locked modem
and instead use a friendly ISP (there are some and there will always be as long as there's
a market for them, i.e. to sell freedom as an extra), get a low end cheap PC running this
specific distro of Linux and get ready to surf and serve.
Now even for locked down modems, if there was one major API for automating forward
connections (such as RSIP mentionned above), it would really help, instead of having
manufacturers' specific solutions. Then it would be easier to force the ISP to provide
hardware that support this API and easier to write software that only needs to deal with
one such API.
Finally, I had this stupid idea one day that it should be possible to hack around
the problem of incoming connections. I never experimented it as I don't really have the
necessary networking knowledge for it. So I'll just mention it here and let the advised
and unadvised reader decide if it's really dumb or if there's some merit to it.
The way I see the problem, two boxes A and B are behind their own NAT, on separate subnets.
They can't talk directly to each others. Instead they can talk to an intermediary box, S,
a server with a fixed IP accessible from the net.
Typically such an intermediary server is used to transit traffic between A and B, or to
enlist dynamic IP repositories. A talks to S and B talks to S but they don't
talk directly to each other because they can't send a packet to a private IP address.
Now A can talk to S thru its NAT server An, and so does B.
S can't initiate a connection to A, it has to come from A. But S can reply too
A by replying with a packet with An's IP and a specific port number, which the NAT
box will recognize and then forward locally to A.
Now the crazy idea here is that B could send those replies to An too by faking them as
coming from S.
Basically the workflow here is:
- A initiates a connection to S thru its NAT, An.
- B initiates a connection to S thru its NAT, Bn.
- S send A the info on how to craft packets for Bn and B how to do it for An.
- B sends packets to An by forging them as replies from S, which An will forward to A.
- A sends packets to Bn by forging them as replies from S, which Bn will forward to B.
I don't have all the specifics of UDP and TCP in mind nor the capacity to really understand them,
nor do I know all the subtleties of NATs and firewall when dealing with incoming connections,
but that's the basic idea. If you think it's crap, please let me know ;-)
Or maybe I should simply patent the idea and let the patent sleep just for the fun of it.
[permalink]
«» 2004/04/20 «»
After reading Tim O'Reilly on Amazon's 1-Click Patent,
I wonder if it would be possible to have GPL Patents, i.e. a software idea
or business logic that is patented using the GPL. Thus the idea would be available to everyone
to use, yet it would have a copyright and not be in the public domain. Moreover, there could
be a requirement that anybody using the idea should explicitely give credit to its original
author (sort of like keeping the original license & author info in a source file).
[permalink]
«» 2004/03/28 «»
Quite interesting:
I can relate to the first one... a lot. The Apple II was my introduction to the computer world.
It truly correspond to the beginning of my real life, somehow. I'm not sure who I was before.
Well actually I know: before I was shy; after I was a shy geek. That makes a huge difference :)
Then of course a major part was all these extra cool demoes on the ST and the Amiga, and the
religious war on which one was better. Already there was the idea that it was getting too easy...
Dunno. For me demoes was a young thing, a teenager/geek mode of expression. I don't quite see
that fit in the adults' world, and most of the demo generation is getting there (or should :)).
Demoes are also a natural product of small but predictable hardware, combined with low
performances which thus require the need for assembly language to get something reasonable out
of it. In today's world, the entry level computer has a powerful click-and-point GUI, non
free development tools (if available at all... in a Windows/Mac world, see Linux below) and
most notably require learning tons of stuff that are pretty high level to do serious programming
(think C++, MFC, Cocoa, vs the ridiculously simple API of the GEM or the 68000 manual).
Luckily, I see two good trends: the growing availability of Linux and Google.
Linux becomes more friendly because it hides its complexity being friendly user interfaces,
but development tools are becoming more and more common too (Glade and GTK+ for example) and
nothing can beat GCC in price vs quality.
Google on the over hand is my first reflex when I need technical information.
The more popular an SDK is, the more chances I have that someone had the same problem and discusses
it on a mailing list. Google will find the info for me. Same for bogus and obscure compiler
messages... Goole will find someone who had the same and the mailing list will probably hold the
reply. Add blogs to mailings lists, it even gets better...
So sure demoes as we knew them are dead.
It's hard for a demo to be really innovative today, and quite frankly an old demo was kind of
like seeing the ultimate real-time computed movie on a computer that could not possibly have
played any kind of movie. Nowaday everything is multimediaesque. A demo is not impressive as
I wouldn't care to execute it when I could see a divx result performing just as well...
Open source dev is really the new thing instead (it's even broader and richer as it is not
limited to one kind of application but basically encompasses everything... except demoes which
were generally close sourced anyway, i.e. free as in free beer, not free speech).
Yet as some other guy was mentionning (hmmm not sure where, a /. comment maybe), engineers who
started with assembly programming generally have a better understanding of the way computers
work -- to which I would add it's even better when one started in electronics engineering ;-)
[permalink]
«» 2004/03/27 «»
Usage of pda...
Cons:
- pda usual app
- graphiti (actually faster to use a virtual keyboard)
- natural writing still inadequate
- not for every day usage
- 320x240 really too small
- palmos 5
Pros:
- as a micro laptop in bed
- wifi
- real keyboard
- web browser
- rss reader
- plucker (-- crashes, no pdf)
- native unicode or japanese
- palmos 6
- .net c# (perfs?)
- need c++ for palmos, can hide os
- app perso: izumi remote editor (not web ui, private protocol or atom)
- blogger
[permalink]
«» 2004/03/19 «»
Very disturbing:
There are many thing wrongs here:
- They are publicly listing IPs of people who downloaded the files, this in my mind is as bad as listing real names. I would not have minded if they had kept stuff anonymous but hey what did they have to publish those IPs? An IP for me is as valid as a name or a street address, that is they are publicly implying someone at this IP did something illegal, which could lead to a lot of trouble.
- By using social engineering to maliciously have their targets run their software, they are using deceptive techniques which are usually used by virus or trojans, so in a way their behavior is as bad as those who create viruses or trokans.
- If this kind of action goes inpunished, it could proliferate, and people who'll mirror this behavior may have way more worse actions in mind. Now this is not exactly the same as diffusing a virus or trojan via a p2p network as this is "legitimate" program that is unlikely to be filtered out by an anti-virus software.
- The two guys behing this are either stupid, or lack any moral values or are simply completly gullible. YMMV.
- These guys are liable for their actions. I don't think anybody will actually sue them, but they would deserve to be sued and at best fined.
Now link that to the previous thought on Doctorow's book being available for free and some being scared by this
as ruining the book industry. Apprently some think P2P is going to turn the world into chaos.
Well too bad for them because you know, it's not going away, and the technology in itself not a bad thing.
File swapping is not exactly something new. It's just more convenient to use a P2P software than swapping
floppies, cds or hard drives as it used to be done in the past.
[permalink]
«» 2004/03/11 «»
Cory Doctorow makes his books available for free under a
Creative Commons license.
This in itself almost justifies supporting this author. At least by reading his material,
and better by owning it.
The bulleting board of Eastern Standard Tribe
is particularly interesting. The second comment is a rant about how allowing free reading of
a book is going to ruin the whole industry, with an emphasis on P2P shared as being thieves.
Of course many object to that shameful argument.
My counter argument is that I already most of my books for free, yet perfectly legally, by
just borrowing them at my local public library (well, it's "fake free" since technically the
library is funded by my taxes and p2p sharing is not 100% free either since I need to buy
the bandwith).
In my life I've also read many books that I borrowed from friends, or that I shared with
other members of my familly.
Does that makes me a thief?
No it does not, yet that's exactly what DRM proponents want to prevent at all cost.
And for books, paperback is still the best medium: compact, mobile, follows me everywhere,
doesn't run out of batteries, can be highlighted, annotated, quoted, teared, makes great gifts
and most important looks really nice on my shelves besides my CD collection.
[permalink]
«» 2004/03/05 «»
Excellent, too many good quotes to put them here :-)
Nice parody here too:
I really like the idea of using these quotes as counter arguments. Unfortunately, I doubt they would enable the
ones using biblical arguments against gay mariages to think any better. They would not use behave like that in the
first place if they hadn't been brainwashed in the first place. Well I hope so.
That on the contrary is not as fun:
[permalink]
«» 2004/02/29 «»
I received this very nice email today:
Bonjour,
Belle randonnée à partir de la Loire Atlantique au fil des émotions échappées de votre site.
Ma fille séjourne depuis peu à Berkeley dans le cadre de ses études.
Je lui souhaite d'ouvrir les yeux sur cet environnement aussi curieusement que vous l'avez
fait afin qu'elle revienne, elle aussi, chargée d'images magnifiques.
Elle a déjà rencontré un "colibri" sur sa route.
Bravo pour votre site et bien cordialement.
Une mère...
This is nice on many levels. I consider it as poetic prose and it's the best compliment I got in a long while.
The page being mentionned is my Ballade à Berkeley.
Yet it yields a more important fact: the most visited part of my web site is the photo album with comments.
I have a new public photo album, RIG based.
It lacks this human touch of having my comments associated with the pictures.
Mostly because RIG doesn't allow that yet (the comment feature is not available yet),
and also because most of the public images are just extracts from a larger set of photos.
So it's one thing to post a big number of photos "as is",
it's another thing to post a web page that uses these same photos to illustrate one's journey.
And that is something no computer tool can do alone. But it can help.
Of course the side effect is that search engines analyze text.
A picture's metadata as read by today's search engines is either a title or a link.
So the more text, the more likely an album page will be visited.
It's something I'll need to think about. Commenting a page like
Ballade à Berkeley is a lot of work, especially since I was doing it
both in English and in French, but the result is rewarding.
[permalink]
«» 2004/02/28 «»
Good reading:
Tried to feed "answer to life the universe and everything"
to the Google Calculator?
Which ends with this message from D. Adams on why 42.
I think it's just a cover story, he's merely trying to mystify the truth :-D.
Of course 42 is the duration of the universe.
Now what we need to figure out is which universe and which time unit.
R
[permalink]
«» 2004/02/21 «»
Outch, that must hurt:
Mixed feelings here:
[permalink]
Blog Archives:
Most recent posts
2004/02/19 - 2003/10/21
[RSS]
This work is licensed by Raphaël Moll under a Creative Commons License.
Display Izumi & PHP Credits
142 accesses, 1 access from 38.107.191.81
Visited 43 times by Google, last 2010/08/17 21:12
Visited 12 times by Yahoo!, last 2009/08/04 01:51
Visited 33 times by MSN, last 2010/07/30 17:51
|
<
Generated in 0.85 seconds the 09/09/2010, 03:38 AM by Izumi 1.1.4 >
|