This text is really the perfect speech to give to the Libertarians, esp the part about their drugs policy. In fact, I can't pull out a single paragraph that I want to take issue with.
Now all I need is for someone to write a nice, lucid essay on how money is not the territory.
Nothing wonderfully exciting today I'm afraid.
Writers for hire by companies and governments. One wonderful quote:"
Will Self said: "I return to the words of Bill Hicks when he said, 'If any artist ever endorses a product then they have completely destroyed their status as an artist.' I don't care if they shit Mona Lisas on cue, they've destroyed their reputations, and advertising for the Government is much more pernicious."
The proceedings of the OLS are up (and have been for a while). A treasure trove of interesting papers contained is contained therein. Highlights (so far) include:
I've finished ripping my CD collection and all they don't make 'em like they used to. My older CDs would be quite happy reading at 4 speed. The newer ones (from about 92 onwards) struggle to manage 1.4 speed (there are a couple in between). The older ones seem to by physically thicker too. I guess they cut down on the quality at some point to reduce production costs.
oh, and mirrors of dvdsynth should they become, ah, required.
Preparations for Imperial continue... Since I don't think I can fit both monitors in my room I've switched to using just one to see if I can manage. It's not too bad, I've had to rejig my desktops and I'm switching much more between them but I think I can cope. The main problem so far is that by bookmarks aren't even close to fitting on one screen. Not using multi-head also means that I can have anti-aliased fonts but I think I need to set them up a little first.
I'm also ripping all my CDs at the moment (OGG, of course). For that I've had to buy a new power supply since my old 230W browned out under the load of all 4 SCSI drives, 2 processors and a DVD drive going. This new one is going fine, hell, I might even spin up the 20GB IDE and the 36GB SCSI that I'm not using at the moment.
The move also means that I'm finally buying all the stuff I should have brought before, namely:
The author of the T0rn rootkit has been arrested [TheReg, BBC] under the Computer Misuse Act. This is a pretty worrying development because, from the sources, it seems that the only `offence' was writing that rootkit, and isn't even very good. Hell, I could do better than that in a couple of days.
Now, I don't support writing rootkits. I know nothing of the accused author, but most of the people using it wouldn't be suitable to wipe shit off my shoe. However, writing it shouldn't be a criminal offence, for two reasons:
Firstly, where you do want to draw the line for `bad software' and who draws it? Is a rootkit bad ("sure, it's only used by little hax0r twits"), so how about exploit code? or fragrouter?, or nmap? or ping -f or DeCSS or even the Linux kernel? If we let the legal system start drawing lines then you just know that we are going to be trapped under a torrent of clueless idiots. That same kind of clueless idiots who are banning all computer games in Greece (I'm afraid that the court decision that said the law was wrong has been overturned) or calling t0rn a "route kit" (I kid you not, on BBC CEEFAX last night).
Secondly, we have a DMCA like "code is speech" argument where you have to draw another line saying "under this is free speech and above it is an illegal tool". The DeCSS case has already shown the futility of that system. Exactly how detailed a description of a rootkit can I write before it's illegal?
Unfortunately it seems that the clueless lawyers have decided to draw these lines anyway. Again. It's gonna be a damm busy wall.
Our very own, AaronSw was on the World Service last night talking about warchalking (right at the end of Newshour). He has links to an ogg (1.8MB) and MP3 (3.2MB). He may well be on NPR's Weekend Edition on Saturday. Go Aaron!
coderman pointed out that I was being an idiot with that map code, it should have been a prefix operator of course.
<coderman> i think c++ makes everyone feel stupid at times. esp. with the STL you get very subtle effects that make sense in hindsight, but are extremely confusing at first light.
Also, coderman suggested that gethrtime would be a good solution to the time problem. Indeed it would, if only it existed in Linux.
This isn't a warblog and, as such, I'm not making any value judgements about the whole Iraq situation. However, I can't stop cracking a smile at the wonderful bait-and-switch that Iraq has pulled. Only 6 days ago, Iraq was saying that inspectors would never be let in (wrapped in a lot of anti-US rhetoric). Now, Bush wants a war (not a value judgement) and saw this as a perfect point of conflict that would bring in the rest of the Security Council. That was the bait and Bush/Blair took it completely saying that Iraq wouldn't be attacked if inspectors went in.
Then yesterday, Iraq switched and listening to the US trying to rebuild their case on Radio4 was just delicious.
When I know what I'm doing I can actually turn out a fair few lines in a day. None of it was anything stunningly deep, but I did about 500 (with some testing and debugging).
Also, I'm going to play about with using weak pointers in this project. Having many interlinked structures (as this code has) can be a real pain when it comes to deleting anything because any dangling pointers left over and pop goes the process.
And this is interesting. To find the highest key in a STL map the second code snippet works and the first prints 0:
printf ("%d\n", (m.end()--)->first);
map<int,int>::iterator i;
i = m.end();
i--;
printf ("%d\n", i->first);
Two things consistently bug me about the kernel, if anyone can send me a solution to either of these I would be most grateful:
Leaving on a more thoughtful note, here's an interesting quote I found reading through the Tao Te Ching (section 38, Stan Rosenthal's translation)
The man who is truly wise and kind
leaves nothing to be done,
but he who only acts
according to his nation's law
leaves many things undone.
Ian has a blog!!
Power cut for about 10 hours today, grumble.
Aaron pointed out that IV's TLS/qmail was probably vulnerable to the OpenSSL bug. I could have sworn that Debian released a security advisory for this, but I couldn't find it and, sure enough, metis still had 0.9.6c. There still isn't a DSA for this, but unstable has 0.9.6g (as does metis now). Thanks Aaron.
JDarcy:
My little corner of the blogosphere seems to have gotten a lot quieter lately. Obviously I've been updating less often than I used to, but many others - e.g. Zooko, AccordionGuy, even Wes Felter - seem to have gone through noticeably fallow periods of late. Whether the result is more or less visible output, everyone seems to be worried about whether they're getting enough (of whatever they want to do) done.
The latest person to catch this apparently-communicable disease is coderman. In his latest article, he laments the slow progress on personal projects, but finds hope in this observation:
I try to keep the posting frequency of IV at a reasonable level, but I do find that I'm not really reading or doing anything wonderfully worthwhile at the moment. In fact all this year I haven't really been coding anything significant. Mostly because I don't have any projects.
People say stuff gets done when programmers have an itch, and it's pretty much true. When I know what I'm doing I code like mad, but I find gumption traps all too common, mainly when something isn't quite right. (I have an unfortunate perfectionist streak). Lately all the itches have been far too big (I've bitched about this before) and I don't feel up to fighting them.
I sometimes think that I should work on Whiterose again since Oskar says that protocol is quite stable now. But then I look at the protocol doc and give up again.
Maybe things will change at Imperial.
(oh, Coderman is being told to spend more time with his wife via her blog )
HashCash isn't a new idea, but it's being talked about again, which is a shame really because I haven't come across a single application where hashcash would work well. Adam Back lists a few at the end of the aforelinked paper, including flood limiting in Freenet. Ignoring the practical problems of integrating hashcash, the major problem is that it scales linearly. If I want to do 1 action, I pay x. If I want to do 5 actions I only pay 5x. There is no way to tell different requesting parties apart, so this is fundamental.
Remember that computers from 5 years ago are going to be about 10 times slower than today's, and you hardly want to cut them off. So you either set the cost far too high, or spammers aren't going to notice it because buying a cheap cluster to calculate hashes isn't really going to bother them. (or even just write a virus/worm to make all the poor Windows users do it for you).
And even in systems where he suggests that hashcash only kick in in a DoS situation (e.g. connection flooding) it doesn't provide "more graceful service" degradation as he claims. It simply moves the bottleneck from the CPU/network to the client, and the fastest client gets served first. (Which would be ok if all the attackers were much slower, but they aren't).
An interesting development would be a computer generatable (my spell checker doesn't like that, but I think it's ok) challenge that only humans could solve. Possibly rendering some text and then messing it up would require a human to solve. That might still be impractical, and spammers could simply hire a sweatshop to solve them all day, but it would be interesting.
Oh, and on the spamming front; that lawyer who got blacklisted wrote back:
When it comes to mail administration, it appears I was several years behind the curve. Since my mail server software, circa 1996, had been purring along quietly without problems since it was new, I had never upgraded it to a version capable of a higher degree of authentication. I'm also old enough to remember when an "open relay" was a relay intentionally left open for anyone to use, not one merely susceptible to misuse. Thanks to all of the readers who wrote to bring me into the new millennium. Both my software and my definition are now upgraded.
At the same time, I labelled the blackhole list operators "vigilantes" for good reason. It was always my understanding that if you lie about your identity to gain access to something that would be closed to you if you told the truth, you've done something wrong. That's true whether you intend to send spam or prevent it. As vile as spam is, the ends don't justify the means. Regardless of whether my mail server used to be "open" or not, I stand by the legal analysis that placed fault on the blackhole operators who forged their identity.
I'm still not sure if this is a spoof or not. If it is, it's a very good one. Quick summary: US lawyer (IP lawyer, naturally) finds his mail server is listed as an open relay, denies that it is one (while giving enough of the story to show that it is) and immediately talks of legal action against the anti-spam group without a thought to fixing the mail server. A good laugh, spoof or (tragically) not.
Via JWZ:
Senator Clinton was booed when she walked on stage last October at a rock concert in Madison Square Garden to benefit 9/11 victims. It was shown live by VH1 but, as ABC's John Stossel illustrated in a July 20/20 special on media distortions, when the Viacom-owned cable channel replayed it sound technicians replaced the booing with cheering and applause. And that version is the permanent record VH1 put onto its DVD of the event.
(all of the following section is tounge-in-cheek )
Installed RH7.3 on a spare drive last night (long story involving odd hardware and a friend needing it) and I'm shocked how easy the install was. Gosh darn it! I can still remember when installing Linux was no mean feat and I'm so young that my first Linux distrib (Slackware) had a 2.0.0 kernel.
In those days an install was a maze of quirks and hardware problems littered with dire warnings about how X would fry your monitor if you got the frequencies wrong. Heck, you were damm lucky if today's kernel managed to exec /bin/sh. In those days spirits were brave, the stakes were high, men were real men, women were real women, and small furry creatures from Alpha Centauri were real small furry creatures from Alpha Centauri.
The damm RH install was graphical (even has a graphical GRUB menu) and picked up all my weird hardware first time and even got X going with DRI. The only thing it didn't detect is that I have 2 monitors. No wonder there are so many Linux lusers on /. if the install is this easy!
There is a proof system for O'Caml called Coq. I keep running off to theme parks and things so I haven't had a chance to read it yet.
Some languages are "safe" in the sense that you cannot dereference NULL pointers etc. Typed languages ensure that arguments to functions cannot be of the wrong type etc. Proven languages can ensure (incomplete list):
In fact you could even give everyone root permissions, but require that programs prove that they aren't doing anything wrong.
This type of thing is obviously very usefully generally but, as has been pointed out many times, we could even get rid of operating systems because no program would do anything nasty (and we could prove this). In a single-level store design having all the programs in a single address space could be a big performance boost.
Mersenne (named after a french monk) primes are of the form 2n-1 where n is an integer, greater then 0. There is a distributed.net like effort to find them called GIMPS (search Google).
A perfect number is a number where the sum of its divisors (excluding itself, but including 1) equals that number. For example 6 is perfect because (1 + 2 + 3) = 6. Thus the sum of all the divisors is twice the number.
Now, I read a while back in a book that (2n-1)(2n-1) was proven to be a perfect number, but the book didn't have the proof. Thankfully, I ran across the proof today. That proof leaves out a number of steps thou, so here's a better one:
One from The Book
After a brief holiday, The Memory Hole is ticking again.
People keep talking about it and it's high time that I looked into it. O'Caml is an ML based language and has all the standard ML language stuff (curried functions etc). It uses inferred typing, which is very useful, despite the few drawbacks (more on that later).
It also has polymorphic typing:
# let f = function (a, b) -> a;; val f : 'a * 'b -> 'a = <fun>
That function takes a 2-tuple of any type and returns the first element and polymorphically works within the type-system.
There is a translation of a French O'Reilly O'Caml book online, but I find it's a little heavy for a introductionary text. I find that this book it nicer to start with. Maybe move on the O'Reilly book afterwards.
This code snippet, which implements red-black binary tree insertion, should demonstrate the power of O'Caml even if you don't understand it. (This assumes you've seen what a mess a red-black insert looks like in C/C++. If not, see this, and I have reasonable reason to suspect there's an error in there since it was done partly from CLR).
let balance = function
Black, Node (Red, Node (Red, a, x, b), y, c), z, d ->
Node (Red, Node (Black, a, x, b), y, Node (Black, c, z, d))
| Black, Node (Red, a, x, Node (Red, b, y, c)), z, d ->
Node (Red, Node (Black, a, x, b), y, Node (Black, c, z, d))
| Black, a, x, Node (Red, Node (Red, b, y, c), z, d) ->
Node (Red, Node (Black, a, x, b), y, Node (Black, c, z, d))
| Black, a, x, Node (Red, b, y, Node (Red, c, z, d)) ->
Node (Red, Node (Black, a, x, b), y, Node (Black, c, z, d))
| a, b, c, d ->
Node (a, b, c, d)
let insert x s =
let rec ins = function
Leaf -> Node (Red, Leaf, x, Leaf)
| Node (color, a, y, b) as s ->
if x < y then balance (color, ins a, y, b)
else if x > y then balance (color, a, y, ins b)
else s
in
match ins s with (* guaranteed to be non-empty *)
Node (_, a, y, b) -> Node (Black, a, y, b)
| Leaf -> raise (Invalid_argument "insert");;
However, there are a couple of silly bits on O'Caml. Firstly, the bitwise (not, logical) AND function is called land. Secondly, the namespace for record fields is flat within a module, so you can't have 2 record/struct types with the same named field in a single module. I'm pretty sure that there isn't a deep reason for that (the shallow reason has to do with type inference).
From one of Coderman's friends....
(Sung to the tune of Y.M.C.A by the Village People) Net geeks There's no need to feel guilt I said, net geeks For the software you built I said, net geeks 'Cause you're not in the wrong There's no need to feel unhappy! Net geeks You can burn a CD I said, net geeks With your fave mp3s You can play them In your home or your car Many ways to take them real far! It's fun to violate the D M C A ! It's fun to violate the D M C A-AY ! You have everything You need to enjoy Your music with your toys! It's fun to violate the D M C A ! It's fun to violate the D M C A-AY ! You can archive your tunes You can share over cable You can annoy the Record Labels!
Photos from the IOI are up. There aren't many, but hopefully Richard got lots more on his (really nice) digital camera.
Discovered a MAP_GROWSDOWN flag in asm/mman.h, unfortunately it doesn't seem to do what one would hope.
Stand have a longish entry on the UK-DMCA. They suggest that the chances of parliament nullifing it are pretty much nil (which, I guess, is depressingly true), and suggest that people write to the UK Patent Office and the Secretary of State for Trade and Industry and try to get more opt-out clauses into the UK law. There has been far too little press coverage about this so far. I've contacted New Scientist, so hopefully they will have something.
There's also a chance that NS will have a short section on the IOI. I only hope they don't include that god awful photo of the team taken at Cambridge. (and, no, I'm not giving the link!)
One of the defining characteristics of quantum physics, over and above the classical, it that it is non-deterministic. Many people have had problems with this, most famously Einstein ("God does not play dice!"). A deterministic universe is very comforting to some people (myself half included) and certainly sits nicely with the logical worlds of maths and computers.
Since many classical processes are statistically modelled (for example, temperature) because it's not useful deal at the level of individual, vibrating molecules some have suggested that the only reason that quantum physics looks non-deterministic is because we are only seeing the cumulative effects of an underlying deterministic system. These ideas are usually called hidden variable theories.
I'm going to run over an argument that TBL gave to me and that's pretty much convinced me to give up on hidden variable theories.
Imagine a light bulb which, at time t, is either on or off and its state is totally random. That's non-deterministic. Not imagine that a pseudo-random number generator (or a Rule 30 CA if you wish, Mr Wolfram) is hidden in the light bulb and actually governs the bulb's state. Given enough time we might be able to reverse the PRNG and that would be a deterministic universe.
Now, our quantum universe might be really non-deterministic or it might have a deterministic process underlying it. However, Bell's Theorem shows that, if there is an underlying deterministic process, then it cannot be localised. So you could not take a section of the universe and have it be deterministic, only (possibly) the whole universe. However, we are in the universe, and so cannot measure every roll of the dice. So in the end you might as well give up on it being deterministic, because it would be a useless determinism anyway.
Well, seems that JDarcy sent in a comment, but changed the subject line so the comment processor rejected it. I thought the subject lines were odd enough looking that people would realise they denote the entry to be commented (or the comment to be commented in the case of threading).
Anyway. The processor (silently) drops malformed mails so we will never know what insightful words JDarcy had the world (unless he reposts).
Having said that, noone has yet managed to get a comment up. People have even mailed me about the comment system rather than post a comment!
Well, turns out the that builders cut the power on Friday (about 3:30 BST) and metis lasted about 45 minutes on the UPS before dying. Since it's softswitched it doesn't power up when the power comes back on (which sucks) and I should get the UPS software working better. Oh, and by the way, I hate builders.
Is an emailed based comments system a blogging first? Well, it's here anyway. Still a little rough about the edges but nothing a little testing (and bug reporting) can't fix (hint!).
Should I put the mails in <pre> tags? At the moment I turn blank lines into <br><br> but that's all.
Reports suggest that E mightn't be the instantly fatal rat-poison we thought it was. Well, we knew that anyway and it's nice to see a report that isn't funded by the US govt to repeat that "Drugs are baaaad".
But don't forget the E is still the most adulterated street drug on the planet. It's been cut with everything under the sun, some of it pretty nasty.
It has a really bad few days for hardware. Firstly, metis (the server which hosts IV) dropped off the face of the net sometime Friday. As I write this it's still down, but I'll be going in to see what died tomorrow. Hopefully it just hit its mean time between failure for the memory and a bad bit flip killed it. Then again, another rat may have pissed in the power supply and shorted it out (I kid you not - that's what killed it last time)
Talking of PSUs I think I need a new one. Twice today my SCSI array failed which causes a reasonably slow, but always fatal, system failure as more and more processes get stuck in the escalator. I'm pretty sure that I'm browning out the power supply having in my case, as I do:
Time for a bigger PSU I think
Ok, so I'm sure it's a really well known result in number theory, but it's the first time I've seen it:
For any integer n, to find the number of factors of the integer find the exponents of the primes in its prime factorisation (call that set a) and eval (reduce #'* (mapcar (lambda (x) (+ x 1)) a)))
For example: where n=12 the prime factorisation is 2^2, 3^1 so a=(2, 1) and it has (2 + 1)(1 + 1) = 6 factors (namely 1,2,3,4,6,12)
It's a pretty standard feature of most blogs that users can add comments. Well, IV has never had that because it would require PHP (or something similar) and I just don't trust apache/PHP to be secure. However, after a remark from Ian I thought that I might be able to have an email based comment system. Watch this space.
I mentioned on Friday that statically analysing untrusted machine code to prove that it's safe to run might be a good idea (go read the post if you haven't already). Well, having thought about it (and looked at RTL) I'm thinking that working with C code might not be so bad after all. I've found a library (CTool) that looks as if it will make the whole parsing a lot easier. Again, watch this space.
BRiX (which I linked to yesterday) popped up on /. while I was in Korea and I've only just got round to having a look at it.
Basically, it's a safe-language OS (where the OS doesn't need to protect processes from each other because the language they are all written in prevents them being bad). It's an old-trick, but I never seen a serious implementation and it would be cool if BRiX reached even a first-beta level (satanic red on black colour scheme not withstanding).
However, I'm undecided on the merits of the safe-language approach. Firstly it mandates that every program be written in the language/byte code (where the changes with each safe-language OS). In BRiX's case the language is called Crush and looks like a typed dialect of Lisp. Learning a new language and rewriting everything in it is a nasty barrier to entry, even if it does hold a certain appeal in terms of cleaning out old code.
With this in mind I'm wondering if a decompiler could statically analyse compiled C code and determine if it's safe. I think, in theory if system calls were treated as non-deterministic, then it should be possible. The practice, on the other hand, might be somewhat painful.
Just a quick FAQ for anybody asking Why work with machine code, why not the C code?
. Mainly because I think I would end up compiling the C code to a reasonably low level anyway. By working from the C code I might get some loop structure etc for free, but since it's perfectly legal to build loops out of gotos I would have to do control-flow analysis anyway, so why have the pain of processing C?
It might be that working with RTL (gcc's intermediate code rep) would be better. There are flags for dumping RTL trees (see your gcc manpage) but I have no experience with RTL. And, of course, if you are happy with the machine code you can run it, but if you are only happy with the RTL you still have to trust a compiler to generate good binaries.
Assorted (but unsorted) links:
Coderman is talking (a little) about P2P radio networks:
Wide spread internetworked wifi hot spots + decentralized peer networks + strong crypto == sweet ass high speed unrestricted digital networks. The possible applications of such networks are extremely exciting (IMHO).
The Freenet team were discussing this idea last year, not really anything to do with Freenet, but as a general point. I don't think this idea is really going to take off until there is a certain density of clueful people with 802.11b in a given area. However, this may already have happened in several places.
Personally, I have far too little experience with 802.11b. The only time I've ever found an AP was in a hotel in Guildford and even then the AP was configured not to do anything. So I could pretty much SNMP walk it, ping it and little else .
However, I agree totally with Coderman that it could be insanely cool.
Much of the work on P2P wireless routing has dealt with getting packets to gateways over a number of hops. Basically a 2 tier network where packets are always going to, or coming from, a gateway. This is a much simpler problem than P2P routing over a network where the nodes will be moving.
Now, one of the cool things that wireless networks do well is broadcast. I mean - it really is a broadcast and the bandwidth needed is independent of the number of nodes that are reached. Most routing protocols are designed for a wired world where broadcasting to n nodes requires n packets. I'm sure there are some cool routing protocols for this which don't require nodes to know their GPS position, thou I'll have to think some more about it.
(and, of course, true broadcast makes DC rings worth considering for some problems)
| / | Root |
| Alternate | The Weird and Wonderful |
| Backlinks | What are backlinks |
| John Gilmore | What's Wrong with Copy Protection |
| Archives | Blog Archives |
| One | Archive 1 |
| Two | Archive 2 |
| Three | Archive 3 |
| Four | Archive 4 |
| Five | Archive 5 |
| Six | Archive 6 |
| Seven | Archive 7 |
| Eight | Archive 8 |
| Nine | Archive 9 |
| Ten | Archive 10 |
| Eleven | Archive 11 |
| Twelve | Archive 12 |
| Thirteen | Archive 13 |
| Fourteen | Archive 14 |
| Fifteen | Archive 15 |
| Sixteen | Archive 16 |
| Seventeen | Archive 17 |
| Eighteen | Archive 18 |
| Nineteen | Archive 19 |
| Twenty | Archive 20 |
| Twenty One | Archive 21 |
| Twenty Two | Archive 22 |
| Twenty Three | Archive 23 |
| Twenty Four | Archive 24 |
| Twenty Five | Archive 25 |
| Twenty Six | Archive 26 |
| Twenty Seven | Archive 27 |
| Twenty Eight | Archive 28 |
| Twenty Nine | Archive 29 |
| Photos | Poor People Caught on Film |
| Jack and the Beanstalk | Jack and the Beanstalk |
| RIP Scan | Results of a Stage Scan Fire |
| Yosemite | Yosemite National Park |
| Projects | Incomplete things from the lab |
| Seagull's Bane | Linux Automounter |
| bttrackd | BitTorrent Tracker |
| CAPTCHA | CAPTCHA CGI script |
| Conserv | Console Serving |
| Deerpark | Using Tor with Firefox/1.1 (Deerpark) |
| DNSFix | Fixing DNS |
| Xovers | XTA Crossover Control |
| IAFS | Archive Org Storage |
| JBIG2 | JBIG2 Encoder |
| Verify | PGP Key Verifier |
| MaxFlow | Maximal Flow in Python |
| PyBloom | Bloom Filters in Python |
| pyGnuTLS | Python wrapping of GnuTLS |
| Sxmap | Apache SuEXEC Map |
| Hellard | Union Server Notes |
| Recordings | Free recordings |
| ICSM Choir | St Paul's Church |
| School | Ancient School Stuff |
| Writings | Who knows |
| Cap Systems | Capability Systems |
| Intro | Introduction to me |
| Suprema | JMC2 Group Project |
| MP Letters | Letters I've written to my MP |
| Sound | Sound With Dramsoc |
| SyncThreading | The wonders of user-land threads |
